sync documentation tree with current master

SConstruct

@@ -254,8 +254,10 @@ def configurePlatform(env):
             problems.append('We need boost::format (header).')
         if not conf.CheckLibWithHeader('boost_program_options-mt','boost/program_options.hpp','C++'):
             problems.append('We need boost::program_options (including binary lib for linking).')
+        if not conf.CheckLibWithHeader('boost_system-mt','boost/system/error_code.hpp','C++'):
+            problems.append('We need the boost::system support library (including binary lib).')
         if not conf.CheckLibWithHeader('boost_filesystem-mt','boost/filesystem.hpp','C++'):
-            problems.append('We need the boost::filesystem (including binary lib for linking).')
+            problems.append('We need the boost::filesystem lib (including binary lib for linking).')
         if not conf.CheckLibWithHeader('boost_regex-mt','boost/regex.hpp','C++'):
             problems.append('We need the boost regular expression lib (incl. binary lib for linking).')
     

src/backend/engine/scheduler-frontend.cpp

@@ -1,8 +1,8 @@
 /*
-  Trafo  -  transforming processing Node 
+  SchedulerFrontend  -  access point to the scheduler within the renderengine
 
   Copyright (C)         Lumiera.org
-    2008,               Hermann Vosseler <Ichthyostega@web.de>
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
 
   This program is free software; you can redistribute it and/or
   modify it under the terms of the GNU General Public License as
@@ -21,13 +21,14 @@
 * *****************************************************/
 
 
-#include "proc/engine/trafo.hpp"
-
-namespace engine
-  {
+#include "backend/engine/scheduler-frontend.hpp"
 
+namespace backend{
+namespace engine {
+  
+  
   /** */
-
-
-
-} // namespace engine
+  
+  
+  
+}} // namespace backend::engine

src/backend/engine/scheduler-frontend.hpp

@@ -0,0 +1,55 @@
+/*
+  SCHEDULER-FRONTEND.hpp  -  access point to the scheduler within the renderengine
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+#ifndef BACKEND_ENGINE_SCHEDULER_FRONTEND_H
+#define BACKEND_ENGINE_SCHEDULER_FRONTEND_H
+
+
+
+//using std::list;
+
+
+namespace backend{
+namespace engine {
+  
+  
+  /**
+   * @todo this is planned to become the frontend
+   * to the render node network, which can be considered
+   * at the lower end of the middle layer; the actual 
+   * render operations are mostly implemented by the backend
+   * ////////TODO WIP as of 12/2010
+   */
+  class SchedulerFrontend
+    {
+    public:
+       ///// TODO: find out about the public operations
+       // note: the play controller lives in the proc-layer,
+       //       but is a subsystem separate of the sesison.
+      
+    private:
+      
+    };
+
+}} // namespace backend::engine
+#endif

src/backend/thread-wrapper.hpp

@@ -200,7 +200,7 @@ namespace backend {
     
     
     /** Synchronisation barrier. In the function executing in this thread
-     *  needs to be a corresponding Thread::sync() call. Blocking until
+     *  needs to be a corresponding Thread::syncPoint() call. Blocking until
      *  both the caller and the thread have reached the barrier.
      */
     void

src/common/guifacade.cpp

@@ -156,7 +156,7 @@ namespace gui {
             if (facade)
               {
                 WARN (guifacade, "GUI subsystem terminates, but GuiFacade isn't properly closed. "
-                               "Closing it forcedly; this indicates broken startup logic and should be fixed.");
+                                 "Closing it forcedly; this indicates broken startup logic and should be fixed.");
                 try { facade.reset (0); }
                 catch(...) { WARN_IF (lumiera_error_peek(), guifacade, "Ignoring error: %s", lumiera_error()); }
                 lumiera_error(); // clear any remaining error state...

src/common/instancehandle.hpp

@@ -184,7 +184,7 @@ namespace lumiera {
       
     public:
       /** Set up an InstanceHandle representing a plugin.
-       *  Should be placed at the client side. 
+       *  Should be placed at the client side.
        *  @param iName unmangled name of the interface
        *  @param version major version
        *  @param minminor minimum acceptable minor version number
@@ -203,7 +203,7 @@ namespace lumiera {
        *  registration and deregistration of interface(s).
        *  Should be placed at the service providing side.
        *  @param a (single) interface descriptor, which can be created with
-       *         LUMIERA_INTERFACE_INSTANCE and referred to by LUMIERA_INTERFACE_REF 
+       *         LUMIERA_INTERFACE_INSTANCE and referred to by LUMIERA_INTERFACE_REF
        */
       InstanceHandle (LumieraInterface descriptor)
         : desc_(descriptor)
@@ -222,9 +222,9 @@ namespace lumiera {
       
       
       
-      /** act as smart pointer providing access through the facade. 
+      /** act as smart pointer providing access through the facade.
        *  @note we don't provide operator*                      */
-      FA * operator-> ()  const { return &(facadeLink_(*this)); }      
+      FA * operator-> ()  const { return &(facadeLink_(*this)); }
       
       /** directly access the instance via the CL interface */
       I& get ()  const { ENSURE(instance_); return *instance_; }

src/common/interface-facade-link.hpp

@@ -0,0 +1,123 @@
+/*
+  INTERFACE-FACADE-LINK  -  a switchable link from interface to service implementation
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+/** @file interface-facade-link.hpp
+ ** Opening, accessing and closing the service access through a facade interface.
+ ** Client code is assumed to access an application level service through an facade
+ ** interface, while the actual implementation object remains an opaque internal detail.
+ ** Moreover, services may come up and shut down, so the implementation might change
+ ** during the Lifecycle. The facility defined here in this header provides a basic
+ ** implementation for this access mechanism, but without any adaptation, binding
+ ** or plugin access layer. It works only under the assumption that both the 
+ ** interface and the actual service implementation coexist in the same
+ ** executable and are written in C++, so any invocation of an 
+ ** interface method boils down to a language-level call.
+ ** 
+ ** Usually, client code doesn't need to include this header. Clients are assumed
+ ** to use the facade interface of the service in question. This facade interface
+ ** contains a static member of type \c lumiera::facade::Accessor<I> (where I is
+ ** the type of the facade interface). The Accessor baseclass is defined in
+ ** interfaceproxy.hpp and typically included through the facade header.
+ ** 
+ ** @note there is a way more elaborate implementation of the same mechanism
+ **       for use with the Lumiera Interface/Plugin system.
+ **
+ ** @see interfaceproxy.hpp description of the more general use case
+ ** @see PlayService example for the simple use case
+ */
+
+
+#ifndef LUMIERA_FACADE_INTERFACE_FACADE_LINK_H
+#define LUMIERA_FACADE_INTERFACE_FACADE_LINK_H
+
+
+#include "lib/error.hpp"
+#include "lib/test/test-helper.hpp"
+#include "include/interfaceproxy.hpp"
+#include "lib/symbol.hpp"
+
+#include <boost/noncopyable.hpp>
+
+
+
+namespace lumiera {
+namespace facade {
+  
+  using lib::Literal;
+  
+  
+  /************************************************************************
+   * simple access-frontend to the implementation of a service (C++ only).
+   * Usually, an instance of Accessor is placed as static member right into
+   * the facade interface used to access the service. This implementation
+   * of the access mechanism handles the simple case that both the facade
+   * and the service implementation are written in C++ and calls happen
+   * within the main executable as direct language calls, without an
+   * binding layer and without involving the Interface/Plugin system.
+   * 
+   * Typically, the InterfaceFacadeLink becomes a member of the service
+   * implementation class and is directly tied into the constructor of
+   * the latter. Being a subclass of lumiera::facade::Accessor, it is
+   * allowed to "open" the facade access just by setting the static
+   * protected pointer Accessor::implProxy_
+   */
+  template<class FA>
+  class InterfaceFacadeLink
+    : protected Accessor<FA>
+    , boost::noncopyable
+    {
+      Literal displayName_;
+      
+      void
+      __checkLifecycle ()
+        {
+          if (Accessor<FA>::implProxy_)
+            throw error::State("Attempt to open an already opened Facade interface."
+                              , error::LUMIERA_ERROR_LIFECYCLE);
+        }
+      
+    public:
+      InterfaceFacadeLink(FA& serviceImpl, Literal interfaceName_for_Log=0)
+        : displayName_(lib::test::showType<FA>(interfaceName_for_Log))
+        {
+          __checkLifecycle();
+          Accessor<FA>::implProxy_ = &serviceImpl;
+          INFO (interface, "interface %s opened", displayName_.c());
+        }
+      
+     ~InterfaceFacadeLink()
+        {
+          INFO (interface, "closing interface %s...", displayName_.c());
+          Accessor<FA>::implProxy_ = 0;
+        }
+    };
+  
+  
+  
+  /** storage for the static access pointer */
+  template<class FA>
+  FA* Accessor<FA>::implProxy_;
+  
+  
+}} // namespace lumiera::facade
+
+#endif

src/common/interfaceproxy.cpp

@@ -28,73 +28,87 @@
 
 using util::cStr;
 
-namespace lumiera {
-  namespace facade {
-    
-    
-    LUMIERA_ERROR_DEFINE (FACADE_LIFECYCLE, "facade interface currently not accessible");  
-    
-    
-    template<class IHA>
-    class Holder;
-    
-    template<class FA, class I>
-    class Holder<InstanceHandle<I,FA> >
-      : Accessor<FA>,
-        protected FA
-      {
-      protected:
-        typedef InstanceHandle<I,FA> IHandle;
-        typedef Holder<IHandle> THolder;
-        typedef Proxy<IHandle> TProxy;
-        typedef Accessor<FA> Access;
-        
-        I& _i_;
-        
-        Holder (IHandle const& iha)
-          : _i_(iha.get())
-          {  }
-        
-      public:
-        static TProxy& open(IHandle const& iha)
-          {
-            static char buff[sizeof(TProxy)];
-            TProxy* p = new(buff) TProxy(iha);
-            Access::implProxy_ = p;
-            return *p;
-          }
-        
-        static void close()
-          {
-            if (!Access::implProxy_) return;
-            TProxy* p = static_cast<TProxy*> (Access::implProxy_);
-            Access::implProxy_ = 0;
-            p->~TProxy();
-          }
-      };
-    
-    
-    template<class FA>
-    FA* Accessor<FA>::implProxy_;
-    
-    
-    template<class IHA>
-    void
-    openProxy (IHA const& iha)
-      {
-        Proxy<IHA>::open(iha);
-      }
-    
-    template<class IHA>
-    void
-    closeProxy ()
-      {
-        Proxy<IHA>::close();
-      }
-    
-   } // namespace facade
+namespace lumiera{
+namespace facade {
   
-} // namespace lumiera
+  
+  LUMIERA_ERROR_DEFINE (FACADE_LIFECYCLE, "facade is closed; service currently not accessible");
+  
+  
+  /**
+   * Implementation Base
+   * for building Facade Proxy implementations.
+   * Typically the purpose of such a proxy is to route
+   * any calls through the C-Bindings of the Lumiera Interface system.
+   * The actual storage for the concrete proxy object is embedded,
+   * inline within the #open() function. For access by the clients,
+   * a frontend-object of type \c Accessor<FA> may be placed into
+   * the facade interface; this accessor-frontend is basically
+   * a concealed static pointer to the proxy, and will be set,
+   * when the interface is opened. This opening and closing
+   * of the interface itself is controlled by the
+   * InstanceHandle, which in turn is typically
+   * created and managed within the context
+   * of the service implementation.
+   */
+  template<class IHA>
+  class Holder;
+  
+  template<class FA, class I>
+  class Holder<InstanceHandle<I,FA> >
+    : Accessor<FA>
+    , protected FA
+    {
+    protected:
+      typedef InstanceHandle<I,FA> IHandle;
+      typedef Holder<IHandle> THolder;
+      typedef Proxy<IHandle> TProxy;
+      typedef Accessor<FA> Access;
+      
+      I& _i_;
+      
+      Holder (IHandle const& iha)
+        : _i_(iha.get())
+        {  }
+      
+    public:
+      static TProxy& open(IHandle const& iha)
+        {
+          static char buff[sizeof(TProxy)];
+          TProxy* p = new(buff) TProxy(iha);
+          Access::implProxy_ = p;
+          return *p;
+        }
+      
+      static void close()
+        {
+          if (!Access::implProxy_) return;
+          TProxy* p = static_cast<TProxy*> (Access::implProxy_);
+          Access::implProxy_ = 0;
+          p->~TProxy();
+        }
+    };
+  
+  
+  template<class FA>
+  FA* Accessor<FA>::implProxy_;
+  
+  
+  template<class IHA>
+  void
+  openProxy (IHA const& iha)
+    {
+      Proxy<IHA>::open(iha);
+    }
+  
+  template<class IHA>
+  void
+  closeProxy ()
+    {
+      Proxy<IHA>::close();
+    }
+  
+}} // namespace lumiera::facade
 
 
 

src/common/subsys.hpp

@@ -104,9 +104,12 @@ namespace lumiera {
       
       
       /** initiate termination of this subsystem.
-       *  may be called repeatedly any time...
-       *  @warning must not block nor throw. */
-      virtual void triggerShutdown ()  throw() =0;
+       *  This trigger may be called repeatedly any time...
+       *  When the subsystem actually has terminated,
+       *  the SigTerm passed to #start must be invoked.
+       * @note called within a locked context (barrier) 
+       * @warning must not block nor throw. */
+      virtual void triggerShutdown ()  throw()  =0;
       
       
       const std::vector<Subsys*>
@@ -114,7 +117,7 @@ namespace lumiera {
       
       
     private:
-      /** weather this subsystem is actually operational.
+      /** whether this subsystem is actually operational.
        *  When returning \c false here, the application may 
        *  terminate at any point without further notice
        *  Note further, that a subsystem must not be in

src/common/subsystem-runner.hpp

@@ -173,7 +173,7 @@ namespace lumiera {
           if (!and_all (susy->getPrerequisites(), isRunning() ))
             {
               susy->triggerShutdown();
-              throw error::Logic("Unable to start all prerequisites of Subsystem "+string(*susy));
+              throw error::State("Unable to start all prerequisites of Subsystem "+string(*susy));
         }   }
       
       void

src/gui/display-service.hpp

@@ -50,6 +50,7 @@
 #include "common/instancehandle.hpp"
 #include "lib/singleton-ref.hpp"
 #include "lib/scoped-ptrvect.hpp"
+#include "include/logging.h"
 
 #include <glibmm.h>
 #include <sigc++/sigc++.h>

src/gui/panels/timeline-panel.cpp

@@ -23,10 +23,12 @@
 
 #include "gui/gtk-lumiera.hpp"
 #include "gui/panels/timeline-panel.hpp"
+#include "gui/widgets/timeline/timeline-zoom-scale.hpp"
 
 #include "gui/workspace/workspace-window.hpp"
 #include "gui/model/project.hpp"
 #include "gui/controller/controller.hpp"
+
 #include "lib/util.hpp"
 
 #include <boost/foreach.hpp>
@@ -34,6 +36,7 @@
 using namespace Gtk;
 using namespace sigc;
 using namespace gui::widgets;
+using namespace gui::widgets::timeline;
 using namespace gui::model;
 
 using boost::shared_ptr;  ///////////////////////////////TICKET #796
@@ -60,6 +63,7 @@ TimelinePanel::TimelinePanel (workspace::PanelManager &panel_manager,
   , iBeamTool(Gtk::StockID("tool_i_beam"))
   , zoomIn(Stock::ZOOM_IN)
   , zoomOut(Stock::ZOOM_OUT)
+  , zoomScale()
   , updatingToolbar(false)
   , currentTool(timeline::Arrow)
 {
@@ -101,14 +105,15 @@ TimelinePanel::TimelinePanel (workspace::PanelManager &panel_manager,
     
   toolbar.append(separator2);
   
-  toolbar.append(zoomIn, mem_fun(this, &TimelinePanel::on_zoom_in));
-  toolbar.append(zoomOut, mem_fun(this, &TimelinePanel::on_zoom_out));
-   
+  toolbar.append(zoomScale);
+  zoomScale.signal_zoom().
+    connect(mem_fun(this,&TimelinePanel::on_zoom));
+
   toolbar.show_all();
   panelBar.pack_start(toolbar, PACK_SHRINK);
 
   // Setup tooltips
-  sequenceChooser     .set_tooltip_text(_("Change sequence"));
+  sequenceChooser .set_tooltip_text(_("Change sequence"));
 
   previousButton  .set_tooltip_text(_("To beginning"));
   rewindButton    .set_tooltip_text(_("Rewind"));
@@ -122,12 +127,18 @@ TimelinePanel::TimelinePanel (workspace::PanelManager &panel_manager,
 
   zoomIn          .set_tooltip_text(_("Zoom in"));
   zoomOut         .set_tooltip_text(_("Zoom out"));
+  zoomScale       .set_tooltip_text(_("Adjust timeline zoom scale"));
 
   // Setup the timeline widget
-  shared_ptr<Sequence> sequence          ///////////////////////////////TICKET #796
+  shared_ptr<Sequence> sequence          ///////////////////////////////TICKET #796 : should use std::tr1::shared_ptr instead of boost
     = *get_project().get_sequences().begin();  
   timelineWidget.reset(new TimelineWidget(load_state(sequence)));
   pack_start(*timelineWidget, PACK_EXPAND_WIDGET);
+
+  // since TimelineWidget is now initialised,
+  // wire the zoom slider to react on timeline state changes
+  zoomScale.wireTimelineState (timelineWidget->get_state(),
+                               timelineWidget->state_changed_signal());
   
   // Set the initial UI state
   update_sequence_chooser();
@@ -136,7 +147,6 @@ TimelinePanel::TimelinePanel (workspace::PanelManager &panel_manager,
   show_time (Time::ZERO);
 }
 
-
 const char*
 TimelinePanel::get_title()
 {
@@ -183,6 +193,13 @@ TimelinePanel::on_ibeam_tool()
   set_tool(timeline::IBeam);
 }
 
+void
+TimelinePanel::on_zoom(double time_scale_ratio)
+{
+  REQUIRE(timelineWidget);
+  timelineWidget->zoom_view(time_scale_ratio);
+}
+
 void
 TimelinePanel::on_zoom_in()
 {
@@ -236,18 +253,19 @@ TimelinePanel::on_sequence_chosen()
     {
       weak_ptr<Sequence> sequence_ptr = 
         (*iter)[sequenceChooserColumns.sequenceColumn];
+
       shared_ptr<Sequence> sequence(sequence_ptr.lock());
+
       if(sequence)
         {
           shared_ptr<timeline::TimelineState> old_state(
             timelineWidget->get_state());
           REQUIRE(old_state);
-            
           if(sequence != old_state->get_sequence())
             timelineWidget->set_state(load_state(sequence));
         }
     }
-    
+
   update_zoom_buttons();
 }
 
@@ -379,9 +397,10 @@ TimelinePanel::on_frame()
 shared_ptr<timeline::TimelineState>
 TimelinePanel::load_state(weak_ptr<Sequence> sequence)
 {
+  /* state exists */
   if(contains(timelineStates, sequence))
-    return timelineStates[sequence];
-  
+      return timelineStates[sequence];
+
   shared_ptr<Sequence> shared_sequence = sequence.lock();
   if(shared_sequence)
     {

src/gui/panels/timeline-panel.hpp

@@ -32,6 +32,8 @@
 #include "gui/panels/panel.hpp"
 #include "gui/widgets/timecode-widget.hpp"
 #include "gui/widgets/timeline-widget.hpp"
+#include "gui/widgets/timeline/timeline-zoom-scale.hpp"
+
 #include "lib/time/timevalue.hpp"
 
 #include <boost/scoped_ptr.hpp>
@@ -87,6 +89,7 @@ private:
   void on_arrow_tool();
   void on_ibeam_tool();
   
+  void on_zoom(double time_scale_ratio);
   void on_zoom_in();
   void on_zoom_out();
   
@@ -171,7 +174,7 @@ private:
   boost::scoped_ptr<TimelineWidget> timelineWidget;
   
   std::map< boost::weak_ptr<model::Sequence>,
-    boost::shared_ptr<widgets::timeline::TimelineState> >          ///////////////////////////////TICKET #796
+    boost::shared_ptr<widgets::timeline::TimelineState> >          ///////////////////////////////TICKET #796 : should use std::tr1::shared_ptr
     timelineStates;
   
   // Toolbar Widgets
@@ -191,6 +194,7 @@ private:
   
   MiniButton zoomIn;
   MiniButton zoomOut;
+  gui::widgets::timeline::TimelineZoomScale zoomScale;
   
   Gtk::SeparatorToolItem separator2;
     

src/gui/widgets/timeline-widget.cpp

@@ -41,7 +41,7 @@ const int TimelineWidget::TrackPadding = 1;
 const int TimelineWidget::HeaderWidth = 150;
 const int TimelineWidget::HeaderIndentWidth = 10;
 const double TimelineWidget::ZoomIncrement = 1.25;
-const int64_t TimelineWidget::MaxScale = 30000000;
+const int64_t TimelineWidget::MaxScale = 30000000; // 30 Million
 
 TimelineWidget::TimelineWidget(
   boost::shared_ptr<timeline::TimelineState> source_state) :
@@ -102,6 +102,7 @@ TimelineWidget::get_state()
 void
 TimelineWidget::set_state(shared_ptr<timeline::TimelineState> new_state)
 { 
+
   state = new_state;
   
   // Clear the track tree
@@ -125,17 +126,17 @@ TimelineWidget::set_state(shared_ptr<timeline::TimelineState> new_state)
   update_tracks();
   
   // Send the state changed signal
-  stateChangedSignal.emit();
+  stateChangedSignal.emit (state);
 }
 
 void
-TimelineWidget::zoom_view(int zoom_size)
+TimelineWidget::zoom_view(double timescale_ratio)
 {
   if(state)
-    {
-      const int view_width = body->get_allocation().get_width();
-      state->get_view_window().zoom_view(view_width / 2, zoom_size);
-    }
+  {
+    const int view_width = body->get_allocation().get_width();
+    state->get_view_window().zoom_view(view_width / 2, timescale_ratio);
+  }
 }
 
 ToolType
@@ -178,7 +179,7 @@ TimelineWidget::hovering_track_changed_signal() const
   return hoveringTrackChangedSignal;
 }
 
-sigc::signal<void>
+TimelineWidget::TimelineStateChangeSignal
 TimelineWidget::state_changed_signal() const
 {
   return stateChangedSignal;
@@ -212,10 +213,12 @@ TimelineWidget::on_view_window_changed()
   if(state)
     { 
       timeline::TimelineViewWindow &window = state->get_view_window();
+
       const int view_width = body->get_allocation().get_width();
-      
+
       horizontalAdjustment.set_page_size(
         window.get_time_scale() * view_width);
+
       horizontalAdjustment.set_value(_raw(window.get_time_offset()));
     }
 }

src/gui/widgets/timeline-widget.hpp

@@ -98,7 +98,7 @@ public:
    * @param zoom_size The number of steps to zoom by. The scale factor
    * is 1.25^(-zoom_size).
    */
-  void zoom_view(int zoom_size);
+  void zoom_view(double timescale_ratio);
   
   /**
    * Gets the type of the tool currently active.
@@ -115,14 +115,16 @@ public:
   
 public:
   /* ===== Signals ===== */
+  typedef sigc::signal<void, boost::shared_ptr<timeline::TimelineState> > TimelineStateChangeSignal;
+  typedef sigc::signal<void, boost::shared_ptr<timeline::Track> > HoveringTrackChangedSignal;
+  
   sigc::signal<void, lib::time::Time> mouse_hover_signal() const;
   
   sigc::signal<void> playback_period_drag_released_signal() const;
   
-  sigc::signal<void, boost::shared_ptr<timeline::Track> >
-    hovering_track_changed_signal() const;
+  HoveringTrackChangedSignal hovering_track_changed_signal() const;
     
-  sigc::signal<void> state_changed_signal() const;
+  TimelineStateChangeSignal state_changed_signal() const;
   
   /* ===== Events ===== */
 protected:
@@ -275,9 +277,8 @@ protected:
   // Signals
   sigc::signal<void, Time> mouseHoverSignal;
   sigc::signal<void> playbackPeriodDragReleasedSignal;
-  sigc::signal<void, boost::shared_ptr<timeline::Track> >
-    hoveringTrackChangedSignal;
-  sigc::signal<void> stateChangedSignal;
+  HoveringTrackChangedSignal hoveringTrackChangedSignal;
+  TimelineStateChangeSignal stateChangedSignal;
     
   bool update_tracks_frozen;
    

src/gui/widgets/timeline/timeline-body.cpp

@@ -39,6 +39,8 @@ using namespace lumiera;
 
 using gui::util::CairoUtil;
 
+using boost::shared_ptr;           ////////////////////TICKET #796
+
 namespace gui {
 namespace widgets {
 namespace timeline {
@@ -61,7 +63,7 @@ TimelineBody::TimelineBody (TimelineWidget &timelineWidget)
   register_styles();
   
   // Reset the state
-  on_state_changed();
+  propagateStateChange();
 }
 
 TimelineBody::~TimelineBody()
@@ -69,6 +71,13 @@ TimelineBody::~TimelineBody()
   WARN_IF(!tool, gui, "An invalid tool pointer is unexpected here");
 }
 
+TimelineViewWindow&
+TimelineBody::viewWindow() const
+{
+  REQUIRE(timelineState);
+  return timelineState->get_view_window();
+}
+
 TimelineWidget&
 TimelineBody::getTimelineWidget () const
 {
@@ -147,7 +156,7 @@ TimelineBody::on_expose_event(GdkEventExpose* event)
   // Makes sure the widget styles have been loaded
   read_styles();
   
-  if(timelineWidget.get_state())
+  if (timelineState)
     {
       // Prepare to render via cairo
       const Allocation allocation = get_allocation();
@@ -169,9 +178,9 @@ TimelineBody::on_scroll_event (GdkEventScroll* event)
 {
   REQUIRE(event != NULL);
   
-  if(timelineWidget.get_state())
+  if (timelineState)
     {
-      TimelineViewWindow &window = view_window();
+      TimelineViewWindow &window = viewWindow();
       const Allocation allocation = get_allocation();
       
       if(event->state & GDK_CONTROL_MASK)
@@ -255,17 +264,16 @@ TimelineBody::on_motion_notify_event(GdkEventMotion *event)
 {
   REQUIRE(event != NULL);
   
-  if(timelineWidget.get_state())
+  if (timelineState)
     {
       // Handle a middle-mouse drag if one is occuring
       switch(dragType)
         {
         case Shift:
           {
-            TimelineViewWindow &window = view_window();
-            
-                                   /////////////////////////////TICKET# 795 : don't reach in from outside and manipulate internals of the timeline view!
+                                   /////////////////////////////TICKET #795 : don't reach in from outside and manipulate internals of the timeline view!
                                    /////////////////////////////            : either encapsulate this entirely here, or leave it to the timeline view!            
+            TimelineViewWindow &window = viewWindow();
             const int64_t scale = window.get_time_scale();
             window.set_time_offset(beginShiftTimeOffset
                                  + TimeValue(scale * (mouseDownX - event->x)));
@@ -294,12 +302,20 @@ TimelineBody::on_motion_notify_event(GdkEventMotion *event)
 }
 
 void
-TimelineBody::on_state_changed()
+TimelineBody::on_state_changed (shared_ptr<TimelineState> newState)
 {
-  if(timelineWidget.get_state())
+  REQUIRE (newState);
+  timelineState = newState;
+  propagateStateChange();
+}
+
+void
+TimelineBody::propagateStateChange()
+{
+  if (timelineState)
     {
       // Connect up some events
-      view_window().changed_signal().connect(
+      viewWindow().changed_signal().connect(
         sigc::mem_fun(this, &TimelineBody::on_update_view) );
     }
     
@@ -373,7 +389,7 @@ TimelineBody::draw_track(Cairo::RefPtr<Cairo::Context> cr,
 
   // Render the track
   cr->save();
-  TimelineViewWindow &window = view_window();
+  TimelineViewWindow &window = viewWindow();
   timeline_track->draw_track(cr, &window);
   cr->restore();
 }
@@ -386,12 +402,11 @@ TimelineBody::draw_selection(Cairo::RefPtr<Cairo::Context> cr)
   // Prepare
   const Allocation allocation = get_allocation();
   
-  shared_ptr<TimelineState> state = timelineWidget.get_state();
-  REQUIRE(state);
+  REQUIRE(timelineState);
   
-  TimelineViewWindow const& window = state->get_view_window();
-  const int start_x = window.time_to_x(state->getSelectionStart());
-  const int end_x   = window.time_to_x(state->getSelectionEnd());
+  TimelineViewWindow const& window = timelineState->get_view_window();
+  const int start_x = window.time_to_x(timelineState->getSelectionStart());
+  const int end_x   = window.time_to_x(timelineState->getSelectionEnd());
   
   // Draw the cover
   if(end_x > 0 && start_x < allocation.get_width())
@@ -427,16 +442,13 @@ TimelineBody::draw_playback_point(Cairo::RefPtr<Cairo::Context> cr)
 {   
   REQUIRE(cr);
   
-  // Prepare
-  
-  shared_ptr<TimelineState> state = timelineWidget.get_state();
-  if(state)
+  if (timelineState)
     {
-      if (!state->isPlaying()) return;
+      if (!timelineState->isPlaying()) return;
       
       const Allocation allocation = get_allocation();
-      Time point = state->getPlaybackPoint();
-      const int x = view_window().time_to_x(point);
+      Time point = timelineState->getPlaybackPoint();
+      const int x = viewWindow().time_to_x(point);
         
       // Set source
       cr->set_source(playbackPointColour);
@@ -455,10 +467,10 @@ TimelineBody::draw_playback_point(Cairo::RefPtr<Cairo::Context> cr)
 void
 TimelineBody::begin_shift_drag()
 {
-  if(timelineWidget.get_state())
+  if (timelineState)
     {
       dragType = Shift;
-      beginShiftTimeOffset = view_window().get_time_offset();
+      beginShiftTimeOffset = viewWindow().get_time_offset();
       beginShiftVerticalOffset = get_vertical_offset();
     }
 }
@@ -475,14 +487,6 @@ TimelineBody::set_vertical_offset(int offset)
   timelineWidget.verticalAdjustment.set_value(offset);
 }
 
-TimelineViewWindow&
-TimelineBody::view_window() const
-{
-  shared_ptr<TimelineState> state = timelineWidget.get_state();
-  REQUIRE(state);
-  return state->get_view_window();
-}
-
 void
 TimelineBody::register_styles() const
 {

src/gui/widgets/timeline/timeline-body.hpp

@@ -68,10 +68,7 @@ public:
    */
   TimelineBody(gui::widgets::TimelineWidget &timeline_widget);
   
-  /**
-   * Destructor
-   */
-  ~TimelineBody();
+  virtual ~TimelineBody();
   
   TimelineWidget&
   getTimelineWidget () const;
@@ -125,13 +122,18 @@ protected:
   bool on_motion_notify_event(GdkEventMotion *event);
   
   /**
-   * The event handler for when the TimelineWidget's state object is
-   * replaced.
+   * The event handler for when the TimelineWidget's state is switched.
    */
-  void on_state_changed();
+  void on_state_changed (boost::shared_ptr<TimelineState> newState);
   
   /* ===== Internals ===== */
 private:
+  /**
+   * Access the current timeline view window
+   * @warning must not be called unless the TimlineWidget
+   *          has a valid state.
+   */
+  TimelineViewWindow& viewWindow() const;
 
   /**
    * Draws the timeline tracks.
@@ -161,13 +163,9 @@ private:
   
   void set_vertical_offset(int offset);
   
-  /**
-   * A helper function to get the view window
-   * @remarks This function must not be called unless the TimlineWidget
-   * has a valid state.
-   */
-  TimelineViewWindow& view_window() const;
-   
+  /** adjust to the new timeline state */
+  void propagateStateChange();
+
   /**
    * Registers all the styles that this class will respond to.
    */
@@ -202,6 +200,8 @@ private:
   Cairo::RefPtr<Cairo::SolidPattern> playbackPointColour;
   
   gui::widgets::TimelineWidget &timelineWidget;
+  boost::shared_ptr<TimelineState> timelineState;      ////////////////////TICKET #796 : should use std::tr1::shared_ptr
+  
 
   friend class Tool;
   friend class ArrowTool;

src/gui/widgets/timeline/timeline-ruler.cpp

@@ -75,6 +75,14 @@ TimelineRuler::TimelineRuler (TimelineWidget &timeline_widget)
   register_styles();
 }
 
+TimelineViewWindow&
+TimelineRuler::viewWindow() const
+{
+  REQUIRE(timelineState);
+  return timelineState->get_view_window();
+}
+
+
 void
 TimelineRuler::set_mouse_chevron_offset(int offset)
 {
@@ -114,7 +122,7 @@ TimelineRuler::on_expose_event(GdkEventExpose* event)
   if(!window)
     return false;
   
-  if(timelineWidget.get_state())
+  if (timelineState)
     {
       // Prepare to render via cairo      
       const Allocation allocation = get_allocation();
@@ -160,11 +168,11 @@ TimelineRuler::on_button_press_event(GdkEventButton* event)
 {
   REQUIRE(event != NULL);
   
-  if(timelineWidget.get_state())
+  if (timelineState)
     {
       if(event->button == 1)
       {
-        pinnedDragTime = view_window().x_to_time(event->x);
+        pinnedDragTime = viewWindow().x_to_time(event->x);
         isDragging = true;
       }
     }
@@ -219,34 +227,40 @@ TimelineRuler::on_size_allocate(Gtk::Allocation& allocation)
 }
 
 void
-TimelineRuler::on_state_changed()
+TimelineRuler::on_state_changed (shared_ptr<TimelineState> newState)
 {
-  if(timelineWidget.get_state())
-    {
-      // Connect up some events
-      view_window().changed_signal().connect(
-        sigc::mem_fun(this, &TimelineRuler::on_update_view) );
-    }
-    
+  REQUIRE (newState);
+  timelineState = newState;
+  
+  propagateStateChange();
+}
+
+void
+TimelineRuler::propagateStateChange()
+{
+  // Connect up some events
+  viewWindow().changed_signal().connect(
+    sigc::mem_fun(this, &TimelineRuler::on_update_view) );
+  
   // Redraw
   on_update_view();
 }
 
+
+
 void
 TimelineRuler::set_leading_x(const int x)
 {
-  shared_ptr<TimelineState> state = timelineWidget.get_state();
-  
-  if(state)
+  if (timelineState)
     {
-      TimeVar newStartPoint (view_window().x_to_time(x));
+      TimeVar newStartPoint (viewWindow().x_to_time(x));
       Offset selectionLength (pinnedDragTime, newStartPoint);
       
       if (newStartPoint > pinnedDragTime)
         newStartPoint=pinnedDragTime; // use the smaller one as selection start
       
-      state->setPlaybackPeriod (Mutation::changeTime(newStartPoint)      );
-      state->setPlaybackPeriod (Mutation::changeDuration(selectionLength));
+      timelineState->setPlaybackPeriod (Mutation::changeTime(newStartPoint)      );
+      timelineState->setPlaybackPeriod (Mutation::changeDuration(selectionLength));
                                    //////////////////////////////////////////////////////TICKET #797 : this is cheesy. Should provide a single Mutation to change all at once
                                                                      ////////////////////TODO        : code duplication with timeline-ibeam-tool 205      
    }
@@ -260,7 +274,7 @@ TimelineRuler::draw_ruler(Cairo::RefPtr<Cairo::Context> cr,
   REQUIRE(ruler_rect.get_width() > 0);
   REQUIRE(ruler_rect.get_height() > 0);
   
-  const TimelineViewWindow &window = view_window();
+  const TimelineViewWindow &window = viewWindow();
   const gavl_time_t left_offset = _raw(window.get_time_offset());
   const int64_t time_scale = window.get_time_scale();
   
@@ -361,16 +375,15 @@ TimelineRuler::draw_selection(Cairo::RefPtr<Cairo::Context> cr,
   REQUIRE(cr);
   REQUIRE(ruler_rect.get_width() > 0);
   REQUIRE(ruler_rect.get_height() > 0);
+  REQUIRE(timelineState);
   
-  shared_ptr<TimelineState> state = timelineWidget.get_state();
-  REQUIRE(state);
-  const TimelineViewWindow &window = state->get_view_window();
+  const TimelineViewWindow &window = timelineState->get_view_window();
 
   Glib::RefPtr<Style> style = get_style();
   Gdk::Cairo::set_source_color(cr, style->get_fg(STATE_NORMAL));
   
   // Draw the selection start chevron
-  const int a = 1 + window.time_to_x(state->getSelectionStart());
+  const int a = 1 + window.time_to_x(timelineState->getSelectionStart());
   if(a >= 0 && a < ruler_rect.get_width())
     {
       cr->move_to(a, ruler_rect.get_height());
@@ -380,7 +393,7 @@ TimelineRuler::draw_selection(Cairo::RefPtr<Cairo::Context> cr,
     }
   
   // Draw the selection end chevron
-  const int b = window.time_to_x(state->getSelectionEnd());
+  const int b = window.time_to_x(timelineState->getSelectionEnd());
   if(b >= 0 && b < ruler_rect.get_width())
     {
       cr->move_to(b, ruler_rect.get_height());
@@ -397,18 +410,17 @@ TimelineRuler::draw_playback_period(Cairo::RefPtr<Cairo::Context> cr,
   REQUIRE(cr);
   REQUIRE(ruler_rect.get_width() > 0);
   REQUIRE(ruler_rect.get_height() > 0);
+  REQUIRE(timelineState);
   
-  shared_ptr<TimelineState> state = timelineWidget.get_state();
-  REQUIRE(state);
-  const TimelineViewWindow &window = state->get_view_window();
+  const TimelineViewWindow &window = timelineState->get_view_window();
   
   // Calculate coordinates
   const float halfSize = playbackPeriodArrowSize / 2;
   
-  const float a = 1.5f + window.time_to_x(state->getPlaybackPeriodStart());
+  const float a = 1.5f + window.time_to_x(timelineState->getPlaybackPeriodStart());
                                     
   const float b = a + halfSize;
-  const float d = 0.5f + window.time_to_x(state->getPlaybackPeriodEnd());
+  const float d = 0.5f + window.time_to_x(timelineState->getPlaybackPeriodEnd());
                              
   const float c = d - halfSize;
   
@@ -465,14 +477,12 @@ TimelineRuler::draw_playback_point(Cairo::RefPtr<Cairo::Context> cr,
   REQUIRE(cr);
   REQUIRE(ruler_rect.get_width() > 0);
   REQUIRE(ruler_rect.get_height() > 0);
+  REQUIRE(timelineState);
   
-  shared_ptr<TimelineState> state = timelineWidget.get_state();
-  REQUIRE(state);
-  if (!state->isPlaying()) return;
+  if (!timelineState->isPlaying()) return;
   
-  TimelineViewWindow const& window = state->get_view_window();
-  Time point = state->getPlaybackPoint();
-  const int x = window.time_to_x(point);
+  Time point = timelineState->getPlaybackPoint();
+  const int x = viewWindow().time_to_x(point);
     
   cr->move_to(x + 0.5, ruler_rect.get_height());
   cr->rel_line_to(0, -playbackPointSize);
@@ -494,7 +504,6 @@ TimelineRuler::calculate_major_spacing() const
 {
   unsigned int i;
   
-  const int64_t time_scale = view_window().get_time_scale();
   const gavl_time_t major_spacings[] = {
       GAVL_TIME_SCALE / 1000,    
       GAVL_TIME_SCALE / 400,
@@ -520,6 +529,8 @@ TimelineRuler::calculate_major_spacing() const
       60ll * 60ll * GAVL_TIME_SCALE
     };     
   
+  const int64_t time_scale = viewWindow().get_time_scale();
+  
   for(i = 0; i < sizeof(major_spacings) / sizeof(gavl_time_t); i++)
     {
       const int64_t division_width = major_spacings[i] / time_scale;
@@ -531,14 +542,6 @@ TimelineRuler::calculate_major_spacing() const
   return major_spacings[i];
 }
 
-TimelineViewWindow&
-TimelineRuler::view_window() const
-{
-  shared_ptr<TimelineState> state = timelineWidget.get_state();
-  REQUIRE(state);
-  return state->get_view_window();
-}
-
 void
 TimelineRuler::register_styles() const
 {

src/gui/widgets/timeline/timeline-ruler.hpp

@@ -29,6 +29,7 @@
 
 #include "gui/gtk-lumiera.hpp"
 #include "lib/time/timevalue.hpp"
+#include "gui/widgets/timeline/timeline-state.hpp"
 
 namespace gui {
 namespace widgets {
@@ -108,10 +109,9 @@ private:
   void on_size_allocate(Gtk::Allocation& allocation);
   
   /**
-   * The event handler for when the TimelineWidget's state object is
-   * replaced.
+   * The event handler for when the TimelineWidget's state is switched.
    */
-  void on_state_changed();
+  void on_state_changed (boost::shared_ptr<TimelineState> newState);         ////////////////////TICKET #796 : should use std::tr1::shared_ptr
   
 private:
   /* ===== Internal Methods ===== */
@@ -161,7 +161,13 @@ private:
    */
   void draw_playback_point(Cairo::RefPtr<Cairo::Context> cr,
     const Gdk::Rectangle ruler_rect);
-
+  
+  /**
+   * After notification of a timeline state switch
+   * do any local adjustments to adapt to the new state 
+   */
+  void propagateStateChange();
+  
   /**
    * Given the current zoom, this function calculates the preiod
    * between major graduations on the ruler scale.
@@ -170,9 +176,9 @@ private:
   gavl_time_t calculate_major_spacing() const;
   
   /**
-   * A helper function to get the view window
+   * Access current timeline view window
    */
-  TimelineViewWindow& view_window() const;
+  TimelineViewWindow& viewWindow() const;
 
   /**
    * Registers all the styles that this class will respond to.
@@ -233,6 +239,11 @@ private:
    */
   gui::widgets::TimelineWidget &timelineWidget;
   
+  /**
+   * the currently active timeline state object 
+   */
+  boost::shared_ptr<TimelineState> timelineState;      ////////////////////TICKET #796 : should use std::tr1::shared_ptr
+  
   /**
    * The caches image of the ruler, over which the chevrons etc. will
    * be drawn.

src/gui/widgets/timeline/timeline-state.cpp

@@ -51,12 +51,13 @@ TimelineState::TimelineState (boost::shared_ptr<model::Sequence> source_sequence
   REQUIRE(sequence);
   
   ////////////////////////////////////////////////////////////TICKET #798: how to handle GUI default state
+  const int64_t DEFAULT_TIMELINE_SCALE =21000000;
   
-  viewWindow.set_time_scale(GAVL_TIME_SCALE / 200);
-  
+  viewWindow.set_time_scale(DEFAULT_TIMELINE_SCALE);
+
   setSelection (Mutation::changeTime (Time(FSecs(2))));
   setSelection (Mutation::changeDuration(Duration(FSecs(2))));
-                               //////////////////////////////////////////////////////TICKET #797 : this is cheesy. Should provide a single Mutation to change all
+  //////////////////////////////////////////////////////TICKET #797 : this is cheesy. Should provide a single Mutation to change all
 }
 
 boost::shared_ptr<model::Sequence>

src/gui/widgets/timeline/timeline-state.hpp

@@ -103,7 +103,7 @@ public:
    *       the GUI is really connected to the Player
    */
   void setPlaybackPoint(Time newPos);
-  
+
   /**
    * A signal to notify when the selected period has changed.
    */
@@ -143,10 +143,9 @@ private:
    *       when actually integrated with the Player
    */
   TimeVar playbackPoint_;
-  
+
   bool isPlayback_;
   
-  
   // Signals
   
   /**

src/gui/widgets/timeline/timeline-view-window.cpp

@@ -31,29 +31,29 @@ namespace gui {
 namespace widgets {
 namespace timeline {
 
-
-
 TimelineViewWindow::TimelineViewWindow (Offset offset, int64_t scale)
   : timeOffset(offset)
   , timeScale(scale)
 {
+  TODO("Create a function to limit timescale between 1 and MaxScale");
+  TODO("TICKET #795 Some functions need to be private");
 }
 
 Offset
-TimelineViewWindow::get_time_offset() const      /////////////////////TODO: this function shouldn't be accessible from outside
+TimelineViewWindow::get_time_offset() const      /////////////////////TICKET #795: this function shouldn't be accessible from outside
 {
   return Offset (timeOffset);
 }
 
 void
-TimelineViewWindow::set_time_offset(TimeValue const& offset) /////////TODO: this function shouldn't be accessible from outside
+TimelineViewWindow::set_time_offset(TimeValue const& offset) /////////TICKET #795: this function shouldn't be accessible from outside
 {
   timeOffset = offset;
   changedSignal.emit();
 }
 
 int64_t
-TimelineViewWindow::get_time_scale() const       /////////////////////TODO: this function shouldn't be accessible from outside
+TimelineViewWindow::get_time_scale() const       /////////////////////TICKET #795: this function shouldn't be accessible from outside
 {
   return timeScale;
 }
@@ -66,21 +66,39 @@ TimelineViewWindow::set_time_scale(int64_t scale)
 }
 
 void
-TimelineViewWindow::zoom_view(int point, int zoom_size)
-{ 
-  int64_t new_time_scale = (double)timeScale * pow(1.25, -zoom_size);
-  
-  // Limit zooming in too close
-  if(new_time_scale < 1) new_time_scale = 1;
-  
-  // Nudge zoom problems caused by integer rounding
-  if(new_time_scale == timeScale && zoom_size < 0)
-    new_time_scale++;
-    
-  // Limit zooming out too far
+TimelineViewWindow::set_time_scale(double ratio)
+{
+  int64_t max = TimelineWidget::MaxScale;
+  int64_t min = 1;
+
+  if(ratio <= 0.0)
+    {
+      set_time_scale((int64_t)min);
+      return;
+    }
+
+  if(ratio > 1.0)
+    {
+      ratio = 1.0;
+    }
+
+   set_time_scale((int64_t)(ratio * max));
+}
+
+void
+TimelineViewWindow::zoom_view(int point, double time_scale_ratio)
+{
+  TODO("Find a Central place for a Zoom Smoothing Factor Variable. Right now it is hard coded at 9.0");
+  int64_t new_time_scale =
+      (int64_t)( pow(time_scale_ratio, 9.0) * (double)TimelineWidget::MaxScale);
+
+  /* Prevent Zooming in To Close and Far */
+  if(new_time_scale < 1)
+    new_time_scale = 1;
+
   if(new_time_scale > TimelineWidget::MaxScale)
     new_time_scale = TimelineWidget::MaxScale;
-  
+
   // The view must be shifted so that the zoom is centred on the cursor
   TimeVar newStartPoint = get_time_offset();
   newStartPoint += TimeValue(point * (timeScale - new_time_scale));

src/gui/widgets/timeline/timeline-view-window.hpp

@@ -94,14 +94,14 @@ public:
    * zero
    */
   void set_time_scale(int64_t time_scale);
-    
+  void set_time_scale(double ratio);
   /**
    * Zooms the view in or out as by a number of steps while keeping a 
    * given point on the timeline still.
-   * @param zoom_size The number of steps to zoom by. The scale factor
-   * is 1.25^(-zoom_size).
+   * @param new_time_scale The number of steps to zoom by. The scale factor
+   * is 1.25^(-new_time_scale).
    */
-  void zoom_view(int point, int zoom_size);
+  void zoom_view(int point, double timescale_ratio);
   
   /**
    * Scrolls the view horizontally as a proportion of the view area.

src/gui/widgets/timeline/timeline-zoom-scale.cpp

@@ -0,0 +1,155 @@
+/*
+  timeline-zoom-scale.cpp  -  Implementation of the zoom scale widget
+
+  Copyright (C)         Lumiera.org
+    2011,               Michael R. Fisher <mfisher31@gmail.com>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+* *****************************************************/
+
+#include "gui/widgets/timeline/timeline-zoom-scale.hpp"
+#include "gui/widgets/timeline-widget.hpp"
+
+using namespace Gtk;
+
+namespace gui {
+namespace widgets {
+
+class TimelineWidget;
+
+namespace timeline {
+
+/**
+ * @todo The initial adjustment value needs to
+ * match what the TimelineViewWindow's actual timeScale
+ * Value is. TimelineViewWindow::get_time_scale() is
+ * currently a public method, but will soon be private.
+ * Maybe TimelineViewWindow can have a zoom_adjustment
+ * that gets passed to this widget's Constructor?
+ * 
+ * @todo actually there is a more involved problem.
+ * The TimelineWidget maintains a TimelineState, which in turn
+ * owns the TimelineViewWindow. Now, the problem is: when we
+ * switch to another Sequence (View), then this TimelineState
+ * gets switched too, causing also a entirely different TimelineViewWindow
+ * to become effective. Thus
+ * - how can we managed to be notified from that switch?
+ * - problem is: TimelineZoomScale widget is owned by the TimelinePannel.
+ *   Likewise, TimelineWidget is owned by the TimelinePannel. But the
+ *   state handling/switching logic is embedded within TimelineWidget
+ * - and finally: how can we translate the actual scale (in time units),
+ *   as maintained within TimelineViewWindow, back into the adjustment
+ *   used here (which uses a relative scale 0...1.0 ) 
+ * 
+ */
+
+TimelineZoomScale::TimelineZoomScale()
+  : HBox()
+  , adjustment(0.5, 0.0, 1.0, 0.000001)
+  , slider()
+  , zoomIn(Stock::ZOOM_IN)
+  , zoomOut(Stock::ZOOM_OUT)
+  , button_step_size(0.03)
+{
+  /* Setup the Slider Control */
+  slider.set_adjustment(adjustment);
+  slider.set_size_request(123,10);
+  slider.set_digits(6);
+  slider.set_inverted(true);
+  slider.set_draw_value(false);
+
+  /* Make our connections */
+  zoomIn.signal_clicked().
+      connect(sigc::mem_fun(this, &TimelineZoomScale::on_zoom_in_clicked));
+
+  zoomOut.signal_clicked().
+      connect(sigc::mem_fun(this, &TimelineZoomScale::on_zoom_out_clicked));
+
+  adjustment.signal_value_changed().
+      connect(sigc::mem_fun(this, &TimelineZoomScale::on_zoom));
+
+  /* Add Our Widgets and show them */
+  pack_start(zoomOut,PACK_SHRINK);
+  pack_start(slider,PACK_SHRINK);
+  pack_start(zoomIn,PACK_SHRINK);
+
+  show_all();
+}
+
+void
+TimelineZoomScale::wireTimelineState (boost::shared_ptr<TimelineState> currentState,
+                                      TimelineWidget::TimelineStateChangeSignal stateChangeSignal)
+{
+  on_timeline_state_changed (currentState);
+  stateChangeSignal.connect (sigc::mem_fun(this, &TimelineZoomScale::on_timeline_state_changed));
+}
+
+void
+TimelineZoomScale::on_timeline_state_changed (boost::shared_ptr<TimelineState> newState)
+{
+  REQUIRE (newState);
+  timelineState = newState;
+  
+  int64_t current_scale =
+      getViewWindow().get_time_scale();
+
+  double linear_scale =
+      (double) current_scale / (double) TimelineWidget::MaxScale;
+
+  /* We have to Revese the Smoothing */
+  TODO("Find a central place for ZoomSmoothingFactor Variable. right now it is 9.0");
+  double new_relative_scale =
+      pow(linear_scale,(1.0/9.0));
+
+  adjustment.set_value(new_relative_scale);
+}
+
+void
+TimelineZoomScale::on_zoom_in_clicked()
+{
+  double newValue = adjustment.get_value() - button_step_size;
+  adjustment.set_value(newValue);
+}
+
+void
+TimelineZoomScale::on_zoom_out_clicked()
+{
+  double newValue = adjustment.get_value() + button_step_size;
+  adjustment.set_value(newValue);
+}
+
+void
+TimelineZoomScale::on_zoom()
+{
+  zoomSignal.emit(adjustment.get_value()) ;
+}
+
+sigc::signal<void, double>
+TimelineZoomScale::signal_zoom()
+{
+  return zoomSignal;
+}
+
+TimelineViewWindow&
+TimelineZoomScale::getViewWindow()
+{
+  REQUIRE (timelineState, "lifecycle error");
+  return timelineState->get_view_window();
+}
+
+} // namespace gui
+} // namespace widgets
+} // namespace timeline

src/gui/widgets/timeline/timeline-zoom-scale.hpp

@@ -0,0 +1,110 @@
+/*
+  timeline-zoom-scale.hpp  -  Declaration of the zoom scale widget
+
+  Copyright (C)         Lumiera.org
+    2011,               Michael R. Fisher <mfisher31@gmail.com>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+/** @file timeline-zoom-scale.hpp
+ ** This file contains the definition of the zoom scale widget
+ */
+
+#ifndef TIMELINE_ZOOM_SCALE_HPP
+#define TIMELINE_ZOOM_SCALE_HPP
+
+#include "gui/gtk-lumiera.hpp"
+#include "gui/widgets/mini-button.hpp"
+#include "gui/widgets/timeline-widget.hpp"
+#include "gui/widgets/timeline/timeline-state.hpp"
+#include "gui/widgets/timeline/timeline-view-window.hpp"
+
+#include <boost/shared_ptr.hpp>
+
+using namespace Gtk;
+using namespace gui::widgets;
+
+namespace gui {
+namespace widgets {
+namespace timeline {
+
+class TimelineZoomScale : public Gtk::HBox
+{
+public:
+  /**
+   * Constructor
+   */
+  TimelineZoomScale();
+
+  /**
+   * Accessor method to the zoomSignal
+   * @return the zoomSignal
+   */
+  sigc::signal<void, double> signal_zoom();
+
+  void wireTimelineState (boost::shared_ptr<TimelineState> currentState,
+                          TimelineWidget::TimelineStateChangeSignal);
+
+private:
+  /* Event Handlers */
+  
+  /**
+   * Update the slider position when the timeline state
+   * is changed.
+   */
+  void on_timeline_state_changed (boost::shared_ptr<TimelineState> newState);         ////////////////////TICKET #796 : should use std::tr1::shared_ptr
+
+  /**
+   * Event handler for when the zoomIn Button
+   * is clicked
+   */
+  void on_zoom_in_clicked();
+
+  /**
+   * Event handler for when the zoomIn Button
+   * is clicked
+   */
+  void on_zoom_out_clicked();
+
+  /**
+   * Event handler for when the adjustment
+   * value is changed
+   */
+  void on_zoom();
+
+  /** access current timeline state */
+  TimelineViewWindow& getViewWindow();
+
+  /* Widgets */
+  Gtk::Adjustment adjustment;
+  Gtk::HScale slider;
+  MiniButton zoomIn;
+  MiniButton zoomOut;
+
+private:
+  /* Signals */
+  sigc::signal<void, double> zoomSignal;
+
+  const double button_step_size;
+
+  boost::shared_ptr<TimelineState> timelineState;
+};
+
+} // namespace gui
+} // namespace widgets
+} // namespace timeline
+
+#endif /* TIMELINE_ZOOM_SCALE_HPP */

src/include/display-facade.h

@@ -21,13 +21,18 @@
 */
 
 /** @file display-facade.h
- ** Major public Interface of the Lumiera GUI. While, generally speaking, the GUI
- ** controls the application and thus acts on its own, it exposes some services
- ** to the lower layers. Especially the lumiera::Display interface serves to
- ** hand over calculated frames to the GUI for displaying them in a viewer.
- ** It's a first draft as of 1/2009, probably it can be factored out into
- ** a more general display service in future.
- **
+ ** Experimental Interface, allowing the Dummy-Player to access the 
+ ** video display widget in the GUI. While, generally speaking, the GUI
+ ** controls the application and thus acts on its own, it might expose some
+ ** services to the lower layers.
+ ** 
+ ** In the Dummy-Player design study, the lumiera::Display interface serves
+ ** to hand over calculated frames to the GUI for displaying them in a viewer.
+ ** 
+ ** This is a first draft as of 1/2009, and likely to be superseded by a
+ ** better design, where rather the \em provider of an output facility
+ ** registers with the OutputManager in the core.
+ ** 
  ** @see gui::GuiFacade
  ** @see dummy-player-facade.h
  ** 

src/include/dummy-player-facade.h

@@ -35,7 +35,6 @@
 #include "include/interfaceproxy.hpp"
 #include "lib/handle.hpp"
 
-#include <boost/noncopyable.hpp>
 
 
 

src/include/interfaceproxy.hpp

@@ -33,7 +33,7 @@
  ** Typically there is another subclass of the Facade interfaces sitting "on the other side"
  ** of the interface barrier and actually implementing the functionality. The template
  ** facade::Accessor can be thought of as a factory creating such a proxy instance of the
- ** facade interface for the client code to use. Typically, in instance of the \em factory
+ ** facade interface for the client code to use. Typically, an instance of the \em factory
  ** is embedded (as a static functor member object) right within the otherwise abstract 
  ** facade interface, this way allowing the client code to write e.g. \c XYZInterface::facade()
  ** to yield a reference to a proxy object implementing \c XYZInterface.
@@ -48,17 +48,18 @@
  ** Any sort of dependency management is outside the scope of the InstanceHandle (for the core
  ** services, it is handled by the dependency of subsystems, while the plugin loader cares
  ** for dependency issues regarding loadable modules, thereby building on the deployment
- ** descriptors.
+ ** descriptors.)
  ** 
  ** For the Layer separation interfaces, the process of loading and opening is abstracted as
  ** an InstanceHandle object. When creating such an InstanceHandle using the appropriate
  ** template and ctor parameters, in addition to the registration with the Interface/Plugin 
- ** system, the corresponding facade::Proxy factory is addressed and "opened" by creating
- ** the right proxy object instance. Similarly, when the InstanceHandle object goes out
- ** of scope, prior to detaching from the Interface/Proxy system, the corresponding 
- ** lumiera::facade::Accessor factory is "closed", which additionally means destroying
- ** the proxy object instance and switching any further access to throwing and exception. 
- **
+ ** system, the corresponding facade::Proxy factory is addressed and the interface instance
+ ** is "opened" by creating the appropriate proxy object instance. Similarly, when the
+ ** InstanceHandle object goes out of scope, prior to detaching from the Interface/Proxy
+ ** system, the corresponding lumiera::facade::Accessor frontend is "closed", which
+ ** additionally means destroying the proxy object instance and switching any
+ ** further access to throwing and exception. 
+ ** 
  ** While client code just includes the interface header (including interfaceproxy.hpp
  ** in turn), there needs to be an actual implementation of each proxy object located in
  ** some translation unit. The usual place is interfaceproxy.cpp, which gets linked into 
@@ -82,45 +83,66 @@
 
 
 namespace lumiera {
-  namespace facade {
-  
-    /** error-ID for accessing a (currently) closed facade */
-    LUMIERA_ERROR_DECLARE(FACADE_LIFECYCLE);  
+namespace facade {
 
+  /** error-ID for accessing a (currently) closed facade */
+  LUMIERA_ERROR_DECLARE(FACADE_LIFECYCLE);  
+
+
+  /*********************************************************************
+   * access-frontend to the implementation of a service.
+   * Usually, an instance of Accessor is placed as static member
+   * right into the facade interface used to access the service.
+   * This allows clients to get the current actual implementation
+   * of that service, just by invoking the function operator on
+   * that member, e.g. \c lumiera::Play::facade()
+   * 
+   * The reason for this rather indirect access technique is Lifecycle:
+   * Service implementations may come up and go down; moreover, a service
+   * might be implemented through a plugin component and thus the actual
+   * invocation needs to be passed through a binding layer. In this case,
+   * clients rather access a proxy object, which then passes on any call
+   * through that binding layer to the actual implementation located
+   * "somewhere".
+   * 
+   * @note the pointer to the actual implementation is a static variable.
+   *       This has two consequences. For one, we're dealing with kind of
+   *       singleton service here. And, secondly, the implementation or
+   *       proxy accessor can inherit from Accessor<FA> where FA is the
+   *       facade interface. Being a subclass, allows the implementation
+   *       to set that pointer when the service comes up, and to clear
+   *       it when the service goes down and the access needs to
+   *       be closed.
+   */
+  template<class FA>
+  class Accessor
+    {
+    protected:
+      static FA* implProxy_;
+      
+      
+    public:
+      FA&
+      operator() ()
+        {
+          if (implProxy_)
+            return *implProxy_;
+          else
+            throw error::State("Facade interface currently closed."
+                              , LUMIERA_ERROR_FACADE_LIFECYCLE);
+        }
+    };
   
-    /*********************************************************************
-     * 
-     */
-    template<class FA>
-    class Accessor
-      {
-      protected:
-        static FA* implProxy_;
-        
-        
-      public:
-        FA&
-        operator() ()
-          {
-            if (implProxy_)
-              return *implProxy_;
-            else
-              throw error::State("Facade interface currently closed.");
-          }
-      };
-    
-    template<class IHA>
-    void openProxy (IHA const&);
-    
-    template<class IHA>
-    void closeProxy ();
-    
-    template<class IHA>
-    class Proxy;
-    
-    
-  } // namespace facade
+  template<class IHA>
+  void openProxy (IHA const&);
   
-} // namespace lumiera
+  template<class IHA>
+  void closeProxy ();
+  
+  template<class IHA>
+  class Proxy;
+  
+  
+}} // namespace lumiera::facade
 
 #endif

src/include/play-facade.h

@@ -0,0 +1,163 @@
+/*
+  PLAYER-FACADE.h  -  access point to the Lumiera player subsystem
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+#ifndef PROC_INTERFACE_PLAY_H
+#define PROC_INTERFACE_PLAY_H
+
+
+
+
+#ifdef __cplusplus  /* ============== C++ Interface ================= */
+
+//#include "include/interfaceproxy.hpp"
+#include "lib/handle.hpp"
+#include "lib/iter-source.hpp"
+#include "lib/time/control.hpp"
+#include "lib/time/timevalue.hpp"
+#include "include/interfaceproxy.hpp"
+#include "proc/mobject/model-port.hpp"
+#include "proc/mobject/output-designation.hpp"
+#include "proc/mobject/session/clip.hpp"
+#include "proc/mobject/session/track.hpp"
+#include "proc/play/output-manager.hpp"
+#include "proc/asset/timeline.hpp"
+#include "proc/asset/viewer.hpp"
+
+
+
+namespace proc {
+namespace play {
+  
+  class PlayProcess;
+}}
+
+
+namespace lumiera {
+  
+  namespace time = lib::time;
+  
+  using std::tr1::weak_ptr;
+    
+    
+    /******************************************************************
+     * Interface to the Player subsystem of Lumiera (Proc-Layer). 
+     * Global access point for starting playback and render processes,
+     * calculating media data by running the render engine.
+     * 
+     * @todo WIP-WIP-WIP 6/2011
+     * @note Lumiera is not yet able actually to deliver rendered data.
+     * @todo there should be an accompanying CL Interface defined for
+     *       the Lumiera interface system, so the player can be 
+     *       accessed from external clients. This was left out
+     *       for now, as we don't have either plugins, nor
+     *       any script running capabilities yet. (5/2011)
+     */
+    class Play
+      {
+      public:
+        
+        /** get an implementation instance of this service */
+        static lumiera::facade::Accessor<Play> facade;
+        
+        
+        /**
+         * Continuous playback process, which has been hooked up
+         * and started with a fixed set of output slots. started with a specific
+         * output size, format and framerate. It is a handle to a calculation process,
+         * which is about to produce a stream of frames and push them to the outputs.
+         * 
+         * The Lifecycle of the referred playback process is managed automatically
+         * through this handle (by ref count). Client code is supposed to use the
+         * API on this handle to navigate and control the playback mode.
+         * 
+         * @see handle.hpp 
+         * @see player-service.cpp implementation
+         */
+        class Controller
+          : public lib::Handle<proc::play::PlayProcess>
+          {
+          public:
+            void play(bool);          ///< play/pause toggle
+            void scrub(bool);         ///< scrubbing playback
+            void adjustSpeed(double); ///< playback speed control
+            void go(lib::time::Time); ///< skip to the given point in time
+            
+            void controlPlayhead (time::Control<time::Time>     & ctrl);
+            void controlDuration (time::Control<time::Duration> & ctrl);
+            void controlLooping  (time::Control<time::TimeSpan> & ctrl);
+            
+            void useProxyMedia (bool);
+            void setQuality (uint);
+            
+            bool is_playing()   const;
+            bool is_scrubbing() const;
+            double getSpeed()   const;
+            uint getQuality()   const;
+            bool usesProxy()    const;
+            
+            operator weak_ptr<proc::play::PlayProcess>()  const;
+          };
+        
+        
+        typedef lib::IterSource<mobject::ModelPort>::iterator    ModelPorts;
+        typedef lib::IterSource<mobject::OutputDesignation>::iterator Pipes;
+        typedef proc::play::POutputManager Output;
+        typedef mobject::session::PClipMO Clip;
+        typedef mobject::PTrack  Track;
+        typedef asset::PTimeline Timeline;
+        typedef asset::PViewer Viewer;
+        
+        /** core operation: create a new playback process
+         *  outputting to the given viewer/display  */
+        virtual Controller connect(ModelPorts, Output)      =0;
+        
+        
+        /* ==== convenience shortcuts for common use cases ==== */
+        Controller perform(Pipes, Output);
+        Controller perform(Timeline);
+        Controller perform(Viewer);
+        Controller perform(Track);
+        Controller perform(Clip);
+        
+      protected:
+        virtual ~Play();
+      };
+    
+    
+} // namespace lumiera
+
+
+
+
+extern "C" {
+#endif /* =========================== CL Interface ===================== */
+
+
+// #include "common/interface.h"
+////////////////////////////////////TODO define a C binding for the Interface system here
+
+
+#ifdef __cplusplus
+}
+#endif
+#endif

src/lib/access-casted.hpp

@@ -34,6 +34,7 @@
  ** e.g. throwing an exception instead of creating a NULL value. 
  ** 
  ** @see lumiera::WrapperPtr usage example to access a variant record
+ ** @see lib::InPlaceAnyHolder usage example to access a subclass in embedded storage
  ** 
  */
 

src/lib/error.hpp

@@ -122,6 +122,7 @@ namespace lumiera {
     LUMIERA_ERROR_DECLARE (ASSERTION);    ///< assertion failure
     
     /* generic error situations */
+    LUMIERA_ERROR_DECLARE (LIFECYCLE);    ///< Lifecycle assumptions violated
     LUMIERA_ERROR_DECLARE (WRONG_TYPE);   ///< runtime type mismatch
     LUMIERA_ERROR_DECLARE (ITER_EXHAUST); ///< end of sequence reached
     LUMIERA_ERROR_DECLARE (BOTTOM_VALUE); ///< invalid or NIL value

src/lib/exception.cpp

@@ -79,6 +79,7 @@ namespace lumiera {
     LUMIERA_ERROR_DEFINE (ASSERTION, "assertion failure");
     
     /* some further generic error situations */
+    LUMIERA_ERROR_DEFINE (LIFECYCLE, "Lifecycle assumptions violated");
     LUMIERA_ERROR_DEFINE (WRONG_TYPE, "runtime type mismatch");
     LUMIERA_ERROR_DEFINE (ITER_EXHAUST, "end of sequence reached");
     LUMIERA_ERROR_DEFINE (BOTTOM_VALUE, "invalid or NIL value");

src/lib/functor-util.hpp

@@ -38,11 +38,15 @@
 #define FUNCTOR_UTIL_H_
 
 #include <tr1/functional>
-#include <cstring>
+#include <boost/functional/hash.hpp>
 
 
 
-namespace util { ////////////TODO: refactor it. But probably not directly into namespace lib. Needs some more consideration though
+namespace lib {
+  typedef size_t HashVal;
+}
+
+namespace util { ////////////TODO: refactor namespace. But probably not directly into namespace lib. Needs some more consideration though
   
   using std::tr1::function;
   using std::tr1::bind;
@@ -86,7 +90,7 @@ namespace util { ////////////TODO: refactor it. But probably not directly into n
   
   /** convenience shortcut to call two functors in sequence. 
    *  @return a Dispatch functor object which incorporates the
-   *          functors as copy and on invocation calls the first 
+   *          functors as copy and on invocation calls the first
    *          function and then returns the result of the second */
   template<typename SIG>
   Dispatch<SIG>
@@ -99,9 +103,12 @@ namespace util { ////////////TODO: refactor it. But probably not directly into n
   
   
   namespace { // hiding some nasty details...
+    
+    using lib::HashVal;
+    using boost::hash_combine;
   
     /**
-     * This Class is used to surpass the access protection
+     * This Class is used to bypass the access protection
      * and break into the tr1::function implementation.
      * Thus we can implement a raw comparison function,
      * as a replacement for the missing functor comparison
@@ -136,6 +143,16 @@ namespace util { ////////////TODO: refactor it. But probably not directly into n
                 && (f1._M_functor._M_unused._M_const_object ==
                     f2._M_functor._M_unused._M_const_object );
           }                      // note: we don't cover any member pointer offset
+        
+        friend HashVal
+        hash_value (HijackedFunction const& fun)
+          {
+            HashVal hash(0);
+            hash_combine (hash, fun.invoker_);
+            hash_combine (hash, fun._M_manager);
+            hash_combine (hash, fun._M_functor._M_unused._M_const_object);
+            return hash;        // note: member pointer offset part uncovered
+          }
       };
     
   }
@@ -184,6 +201,41 @@ namespace util { ////////////TODO: refactor it. But probably not directly into n
   }
   
   
+  /** workaround to calculate a hash value for a given function object. 
+   * @note use with caution. This implementation relies on internal details
+   *       of boost/function; it can be expected to be rather conservative,
+   *       i.e. yielding different hash values for objects, which actually
+   *       are semantically equivalent.
+   * @warning especially function objects bound to member functions aren't
+   *       fully supported. It \em may happen that we miss differences on the
+   *       offset part and only hash the "this" pointer on some platform.
+   */
+  template<typename SIG>
+  inline HashVal
+  rawHashValue (function<SIG> const& fun)
+  {
+    typedef HijackedFunction const& Hij;
+    
+    return hash_value (reinterpret_cast<Hij> (fun));
+  }
+  
 } // namespace util
 
+namespace std {
+namespace tr1 {
+  
+  /** inject into std::tr1 to be picked up by ADL:
+   * @return hash value of given functor
+   * @note   use with caution. Hash is calculated
+   *         relying on undocumented boost internals.
+   */
+  template<typename SIG>
+  inline lib::HashVal
+  hash_value (function<SIG> const& fun)
+  {
+    return util::rawHashValue (fun);
+  }
+  
+}}
+
 #endif /*UTIL_HPP_*/

src/lib/handle.hpp

@@ -60,7 +60,7 @@ namespace lib {
    * Generic opaque reference counting handle, for accessing a service
    * and managing its lifecycle. Usually such a handle is created by
    * an service interface and \link #activate activated \endlink by
-   * setting up the link to some internal implementation object.
+   * setting up the link to a suitable hidden implementation object.
    * This setup can only be done by a friend or derived class,       //////////////////////////TODO: that was the intention. Why didn't this work out as expected?
    * while client code is free to copy and store handle objects.
    * Finally, any handle can be closed, thereby decrementing

src/lib/iter-source.hpp

@@ -279,6 +279,14 @@ namespace lib {
           typedef typename IterSource<Val>::iterator Iter;
         };
       
+      template<class IT, class FUN>
+      struct _TransformIterT
+        {
+          typedef typename _ProducedOutput<FUN>::Type ResVal;          
+          typedef TransformIter<IT,ResVal>         TransIter;
+          typedef typename IterSource<ResVal>::iterator Iter;
+        };
+      
       template<class IT>
       struct _PairIterT
         {
@@ -329,6 +337,28 @@ namespace lib {
     }
     
     
+    /** pipes a given Lumiera Forward Iterator through
+     *  a transformation function and wraps the resulting
+     *  transforming Iterator, exposing just a IterSource.
+     *  This convenience shortcut can be used to build a
+     *  processing chain; the resulting IterSource will
+     *  hide any involved detail types.
+     * @note as with any IterSource, there is one virtual
+     *       function call for every fetched element.
+     */
+    template<class IT, class FUN>
+    typename _TransformIterT<IT,FUN>::Iter
+    transform (IT const& source, FUN processingFunc)
+    {
+      typedef typename _TransformIterT<IT,FUN>::ResVal    ValType;
+      typedef typename _TransformIterT<IT,FUN>::TransIter TransIT;
+      
+      return IterSource<ValType>::build (
+          new WrappedLumieraIterator<TransIT> (
+              transformIterator (source, processingFunc)));
+    }
+    
+    
     /** @return a Lumiera Forward Iterator to yield
      *          all the keys of the given Map or Hashtable
      */
@@ -424,6 +454,7 @@ namespace lib {
 
   }
   using iter_source::wrapIter;
+  using iter_source::transform;
   using iter_source::eachMapKey;
   using iter_source::eachDistinctKey;
   using iter_source::eachValForKey;

src/lib/lifecycle.cpp

@@ -76,11 +76,11 @@ namespace lumiera {
 extern "C" { /* ==== implementation C interface for lifecycle hooks ======= */
   
   
-  extern const char * lumiera_ON_BASIC_INIT       = lumiera::ON_BASIC_INIT;
-  extern const char * lumiera_ON_GLOBAL_INIT      = lumiera::ON_GLOBAL_INIT;
-  extern const char * lumiera_ON_GLOBAL_SHUTDOWN  = lumiera::ON_GLOBAL_SHUTDOWN;
+  const char * lumiera_ON_BASIC_INIT       = lumiera::ON_BASIC_INIT;
+  const char * lumiera_ON_GLOBAL_INIT      = lumiera::ON_GLOBAL_INIT;
+  const char * lumiera_ON_GLOBAL_SHUTDOWN  = lumiera::ON_GLOBAL_SHUTDOWN;
   
-  extern const char * lumiera_ON_EMERGENCY        = lumiera::ON_EMERGENCY;
+  const char * lumiera_ON_EMERGENCY        = lumiera::ON_EMERGENCY;
   
   
   

src/lib/meta/function-erasure.hpp

@@ -22,6 +22,7 @@
 
 
 /** @file function-erasure.hpp
+ ** Generic holder for functor objects, concealing the concrete function type.
  ** When working with generic function objects and function pointers typed to
  ** arbitrary signatures, often there is the necessity to hold onto such a functor
  ** while hiding the actual signature behind an common interface ("type erasure").

src/lib/meta/generator-combinations.hpp

@@ -0,0 +1,101 @@
+/*
+  GENERATOR-COMBINATIONS.hpp  -  generate combinations and variations
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+/** @file generator-combinations.hpp
+ ** Metaprogramming facilities to generate combination cases.
+ ** Similar to the plain typelist-based generators, a custom supplied
+ ** template will be instantiated with combinations of the parameter types
+ ** and then mixed into the resulting type
+ ** 
+ ** @see generator-combinations-test.cpp
+ ** @see generator.hpp
+ ** 
+ */
+
+
+#ifndef LUMIERA_META_GENERATOR_COMBINATIONS_H
+#define LUMIERA_META_GENERATOR_COMBINATIONS_H
+
+#include "lib/meta/typelist.hpp"
+#include "lib/meta/typelist-manip.hpp"
+#include "lib/meta/generator.hpp"
+
+
+
+namespace lumiera {
+namespace typelist{
+  
+  
+  template<class TYPES_1, class TYPES_2>
+  struct CartesianProduct
+    : Distribute< typename TYPES_1::List
+                , typename TYPES_2::List
+                >
+    { };
+  
+  
+  
+  template< template<class,class,class> class _X_>
+  struct PickParametersFromSublist
+    {
+      template<class SUBLIST, class BASE>
+      struct SingleCaseInstantiation
+        : _X_< typename Pick<SUBLIST,0>::Type
+             , typename Pick<SUBLIST,1>::Type
+             , BASE
+             >
+        { };
+    };
+  
+  
+  /**
+   * Build a Case matrix.
+   * The given parameter template _X_
+   * will be instantiated for each possible combination
+   * of the elements from both parameter type-lists.
+   * All these instantiations will be chained up
+   * into a linear inheritance chain rooted
+   * at the BASE type.
+   * @note the custom-supplied template _X_ needs to take a 3rd parameter,
+   *       and inherit from this parameter, in order to form that chain.
+   *       Typically you'll define some (static) functions within that
+   *       template, which then forward the call to the given BASE
+   *       (and of course, that BASE then needs to define this
+   *       function as well). 
+   */
+  template
+    < class TYPES_1, class TYPES_2           ///< the two type collections to pick combinations from
+    , template<class,class,class> class _X_  ///< template with two arg types and a base type
+    , class BASE = NullType
+    >
+  struct InstantiateChainedCombinations
+    : InstantiateChained< typename CartesianProduct<TYPES_1,TYPES_2>::List
+                        , PickParametersFromSublist<_X_>::template SingleCaseInstantiation
+                        , BASE
+                        >
+    { };
+  
+  
+  
+}} // namespace lumiera::typelist
+#endif

src/lib/meta/typelist-manip.hpp

@@ -0,0 +1,329 @@
+/*
+  TYPELIST-MANIP.hpp  -  Utils for working with lists-of-types
+
+  Copyright (C)         Lumiera.org
+    2008,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+/** @file typelist-manip.hpp
+ ** Metaprogramming: Helpers for manipulating lists-of-types.
+ ** Sometimes, we use metaprogramming to generate a variation of concrete
+ ** implementations by combining some basic building blocks. Typically, there
+ ** is a number of similar, but not suitably related types involved. We want to
+ ** process those types using a common scheme, without being forced to squeeze
+ ** all those types into a artificial inheritance relationship. Now, generating
+ ** some kind of common factory or adapter, while mixing in pieces of code tailored
+ ** specifically to the individual types, allows still to build a common processing
+ ** in such situations.
+ ** 
+ ** The facilities in this header provide the basics of simple functional list
+ ** processing (mostly with tail recursion). Usually, there is one template parameter
+ ** TYPES, which accepts a \em Type-list. The result of the processing step is then
+ ** accessible as an embedded typedef named \c List . Here, all of the 'processing'
+ ** to calculate this result is performed by the compiler, as a side-effect of
+ ** figuring out the resulting concrete type. At run time, in the generated
+ ** code, typically the resulting classes are empty, maybe just
+ ** exposing a specifically built-up function. 
+ ** 
+ ** @see generator.hpp
+ ** @see typelist-manip-test.cpp
+ ** @see TimeControl_test usage example
+ ** @see typelist.hpp
+ ** 
+ */
+
+
+#ifndef LUMIERA_META_TYPELIST_MANIP_H
+#define LUMIERA_META_TYPELIST_MANIP_H
+
+
+
+#include "lib/meta/typelist.hpp"
+
+namespace lumiera {
+namespace typelist{
+    
+    
+    /** pick the n-th element from a typelist */
+    template<class TYPES, uint i>
+    struct Pick
+      {
+        typedef NullType Type;  
+      };
+    template<class TY, class TYPES>
+    struct Pick<Node<TY,TYPES>, 0>
+      {
+        typedef TY Type;  
+      };
+    template<class TY, class TYPES, uint i>
+    struct Pick<Node<TY,TYPES>, i>
+      {
+        typedef typename Pick<TYPES, i-1>::Type Type;
+      };
+    
+    
+    
+    
+    /** apply a transformation (template) to each type in the list */
+    template<class TY, template<class> class _TRANS_>
+    struct Apply                           { typedef TY List; };
+    
+    template< class TY, class TYPES
+            , template<class> class _TRANS_
+            >
+    struct Apply<Node<TY,TYPES>, _TRANS_ > { typedef Node< typename _TRANS_<TY>::Type
+                                                         , typename Apply<TYPES,_TRANS_>::List
+                                                         > List; };
+    
+    
+    /** conditional node: skip an element based on evaluating a predicate */
+    template<bool, class T, class TAIL>
+    struct CondNode                        { typedef TAIL  Next; };
+    
+    template<class T, class TAIL>
+    struct CondNode<true, T, TAIL>         { typedef Node<T,TAIL>  Next; };
+    
+    /** filter away those types which don't fulfil a predicate metafunction */
+    template< class TYPES 
+            , template<class> class _P_    ///< a template providing a boolean member \c ::value
+            >
+    struct Filter;
+    
+    template<template<class> class _P_>
+    struct Filter<NullType,_P_>            { typedef NullType  List; };
+    
+    template< class TY, class TYPES
+            , template<class> class _P_
+            >
+    struct Filter<Node<TY,TYPES>,_P_>      { typedef typename CondNode< _P_<TY>::value
+                                                                      , TY
+                                                                      , typename Filter<TYPES,_P_>::List
+                                                                      >::Next      
+                                                                      List; };
+    
+    
+    /** append lists-of-types */
+    template<class TY1, class TY2>
+    struct Append                          { typedef Node<TY1, typename Append<TY2,NullType>::List>  List; };
+    
+    template< class TY, class TYPES
+            , class TAIL
+            >
+    struct Append<Node<TY,TYPES>, TAIL>    { typedef Node<TY,  typename Append<TYPES,  TAIL>::List>  List; };
+    
+    template<class TY, class TYPES>
+    struct Append<NullType, Node<TY,TYPES> >   { typedef Node<TY,TYPES>   List; };
+    
+    template<class TY, class TYPES>
+    struct Append<Node<TY,TYPES>, NullType>    { typedef Node<TY,TYPES>   List; };
+    
+    template<class TY1>
+    struct Append<TY1,NullType>            { typedef Node<TY1,NullType>   List; };
+    
+    template<class TY2>
+    struct Append<NullType,TY2>            { typedef Node<TY2,NullType>   List; };
+    
+    template<>
+    struct Append<NullType,NullType>       { typedef NullType             List; };
+    
+    
+    
+    
+    /** access the last list element */
+    template<class TYPES>
+    struct SplitLast;
+    
+    template<>
+    struct SplitLast<NullType>             { typedef NullType Type;
+                                             typedef NullType List; };
+    template<class TY>
+    struct SplitLast<Node<TY,NullType> >   { typedef TY       Type;
+                                             typedef NullType List; };
+    
+    template<class TY, class TYPES>
+    struct SplitLast<Node<TY,TYPES> >      { typedef typename SplitLast<TYPES>::Type Type;
+                                             typedef typename Append< TY,
+                                                                      typename SplitLast<TYPES>::List
+                                                                    >::List 
+                                                                    List; };
+    
+    
+    
+    /** 
+     * splice a typelist like an overlay
+     * into an base typelist, starting at given index.
+     * @return either the combined (spliced) List, or
+     *         the Front/Back part before or after the Overlay
+     * @note using a NullType as OVERLAY allows to extract
+     *         an arbitrary Front/Back part of the list
+     */
+    template<class BASE, class OVERLAY, uint i=0>
+    struct Splice;
+    
+    template<class B, class BS,
+             class OVERLAY, uint i>
+    struct Splice<Node<B,BS>, OVERLAY, i>  { typedef Node<B, typename Splice<BS, OVERLAY, i-1>::List>  List;
+                                             typedef Node<B, typename Splice<BS, OVERLAY, i-1>::Front> Front;
+                                             typedef         typename Splice<BS, OVERLAY, i-1>::Back   Back; };
+    
+    template<class B, class BS,
+             class O, class OS >
+    struct Splice<Node<B,BS>,Node<O,OS>,0> { typedef Node<O, typename Splice<BS,OS, 0>::List>          List;
+                                             typedef NullType                                          Front;
+                                             typedef         typename Splice<BS,OS, 0>::Back           Back; };
+    
+    template<class B, class BS>
+    struct Splice<Node<B,BS>, NullType, 0> { typedef Node<B, BS> List;
+                                             typedef NullType    Front;
+                                             typedef Node<B, BS> Back; };
+    
+    template<class XX, uint i>
+    struct Splice<NullType, XX, i>         { typedef NullType    List;
+                                             typedef NullType    Front;
+                                             typedef NullType    Back; };
+    
+    
+    
+    
+    /**
+     * Allows to access various parts of a given typelist:
+     * Start and End, Prefix and Tail...
+     */
+    template<class TYPES>
+    struct Dissect;
+    
+    template<class T, class TYPES>
+    struct Dissect<Node<T,TYPES> >
+      {
+        typedef Node<T,TYPES>                  List;  ///< the complete list
+        typedef T                              Head;  ///< first element
+        typedef Node<T,NullType>               First; ///< a list containing the first element
+        typedef TYPES                          Tail;  ///< remainder of the list starting with the second elm.
+        typedef typename SplitLast<List>::List Prefix;///< all of the list, up to but excluding the last element
+        typedef typename SplitLast<List>::Type End;   ///< the last element
+        typedef Node<End,NullType>             Last;  ///< a list containing the last element
+      };
+    
+    template<>
+    struct Dissect<NullType>
+      {
+        typedef NullType                       List;
+        typedef NullType                       Head;
+        typedef NullType                       First;
+        typedef NullType                       Tail;
+        typedef NullType                       Prefix;
+        typedef NullType                       End;
+        typedef NullType                       Last;
+      };
+    
+    
+    
+    
+    /** 
+     * prefix each of the elements,
+     * yielding a list-of lists-of-types
+     */
+    template<class T, class TY>
+    struct PrefixAll                       { typedef Node< typename Append<T,TY>::List, NullType>  List; };
+    
+    template<class T>
+    struct PrefixAll<T, NullType>          { typedef NullType  List; };
+    
+    template<class T>
+    struct PrefixAll<T, NodeNull>          { typedef Node< typename Append<T,NodeNull>::List, NullType>  List; };
+    
+    template< class T
+            , class TY, class TYPES
+            >
+    struct PrefixAll<T, Node<TY,TYPES> >   { typedef Node< typename Append<T,TY>::List
+                                                         , typename PrefixAll<T,TYPES>::List
+                                                         >     List; };
+    
+    
+    
+    
+    /**
+     * build a list-of lists, where each element of the first arg list
+     * gets in turn prepended to all elements of the second arg list.
+     * Can be used to build all possible combinations from two
+     * sources, i.e. the Cartesian product.
+     */
+    template<class TY1,class TY2>
+    struct Distribute                      { typedef typename PrefixAll<TY1,TY2>::List  List; };
+    
+    template<class TY>
+    struct Distribute<NullType,TY>         { typedef NullType List; };
+    
+    template< class TY, class TYPES
+            , class TAIL
+            >
+    struct Distribute<Node<TY,TYPES>,TAIL> { typedef typename Append< typename PrefixAll<TY,TAIL>::List
+                                                                    , typename Distribute<TYPES,TAIL>::List
+                                                                    >::List                    
+                                                                    List; };
+    
+    
+    
+    /** 
+     * build all possible combinations, based on a enumeration of the basic cases.
+     * For each of the types in the argument list, an "enumeration generator" template is invoked,
+     * yielding a list of the possible base cases. These base cases are then combined with all the
+     * combinations of the rest, yielding all ordered combinations of all cases. Here, "ordered"
+     * means that the base cases of the n-th element will appear in the n-th position of the
+     * resulting lists,
+     * 
+     * For the typical example, the "base cases" are {flag(on), flag(off)}, so we get a
+     * list-of-lists, featuring all possibilities to combine these distinct toggles. 
+     */
+    template< class X
+            , template<class> class _ENUM_>
+    struct Combine                         { typedef typename Distribute< typename _ENUM_<X>::List
+                                                                        , Node<NullType,NullType>
+                                                                        >::List  List; };
+    template< template<class> class _ENUM_>
+    struct Combine<NullType, _ENUM_ >      { typedef NodeNull                    List; };
+    
+    template< class TY, class TYPES
+            , template<class> class _ENUM_>
+    struct Combine<Node<TY,TYPES>,_ENUM_>  { typedef typename Distribute< typename _ENUM_<TY>::List
+                                                                        , typename Combine<TYPES,_ENUM_>::List
+                                                                        >::List  List; };
+    
+    /** enumeration generator for the Combine metafunction,
+     *  yielding an "on" and "off" case
+     */
+    template<class F>
+    struct FlagOnOff
+      { 
+        typedef Node<F, Node<NullType,NullType> >  List;
+      };
+    
+    
+    /** generate all possible on-off combinations of the given flags */
+    template<class FLAGS>
+    struct CombineFlags
+      { 
+        typedef typename Combine<FLAGS, FlagOnOff>::List  List;
+      };
+    
+    
+    
+    
+}} // namespace lumiera::typelist
+#endif

src/lib/meta/typelist-util.hpp

@@ -1,5 +1,5 @@
 /*
-  TYPELIST-UTIL.hpp  -  Utils for working with lists-of-types
+  TYPELIST-UTIL.hpp  -  simple helpers for working with lists-of-types
 
   Copyright (C)         Lumiera.org
     2008,               Hermann Vosseler <Ichthyostega@web.de>
@@ -21,11 +21,32 @@
 */
 
 
+/** @file typelist-util.hpp
+ ** Metaprogramming: simple helpers for working with lists-of-types. 
+ ** This header provides some very basic "meta functions" for extracting
+ ** some informations from a list-of-types. In Lumiera, we use template
+ ** metaprogramming and especially such lists-of-types, whenever we build
+ ** some common implementation backbone, without being able to subsume all
+ ** participating types (classes) into a single inheritance hierarchy.
+ ** 
+ ** The "meta functions" defined here are templates; to access the "result" of
+ ** such a meta function, we instantiate the template and then access one of the
+ ** embedded constant definitions (usually the enum constant named \c value)
+ ** 
+ ** @see generator.hpp
+ ** @see TypelistUtil_test
+ ** @see lib::SimpleAllocator usage example (for isInList)
+ ** @see TypelistManip_test
+ ** @see typelist.hpp
+ ** 
+ */
+
+
 #ifndef LUMIERA_META_TYPELIST_UTIL_H
 #define LUMIERA_META_TYPELIST_UTIL_H
 
 
-  
+
 #include "lib/meta/typelist.hpp"
 
 namespace lumiera {
@@ -34,6 +55,7 @@ namespace typelist{
     
     /**
      * Metafunction counting the number of Types in the collection
+     * @return an embedded constant \c value holding the result
      */
     template<class TYPES>
     struct count;
@@ -48,6 +70,7 @@ namespace typelist{
         enum{ value = 1 + count<TYPES>::value };
       };
     
+    
     /**
      * Metafunction " max( sizeof(T) ) for T in TYPES "
      */
@@ -68,233 +91,35 @@ namespace typelist{
       };
     
     
-    /** apply a transformation (template) to each type in the list */
-    template<class TY, template<class> class _TRANS_>
-    struct Apply                           { typedef TY List; };
-    
-    template< class TY, class TYPES
-            , template<class> class _TRANS_
-            >
-    struct Apply<Node<TY,TYPES>, _TRANS_ > { typedef Node< typename _TRANS_<TY>::Type
-                                                         , typename Apply<TYPES,_TRANS_>::List
-                                                         > List; };
-    
-    
-    template<bool, class T, class TAIL>
-    struct CondNode                        { typedef TAIL  Next; };
-    
-    template<class T, class TAIL>
-    struct CondNode<true, T, TAIL>         { typedef Node<T,TAIL>  Next; };
-    
-    /** filter away those types which don't fulfil a predicate metafunction */
-    template< class TYPES 
-            , template<class> class _P_    ///< a template providing a boolean member \c ::value
-            >
-    struct Filter;
-    
-    template<template<class> class _P_>
-    struct Filter<NullType,_P_>            { typedef NullType  List; };
-    
-    template< class TY, class TYPES
-            , template<class> class _P_
-            >
-    struct Filter<Node<TY,TYPES>,_P_>      { typedef typename CondNode< _P_<TY>::value
-                                                                      , TY
-                                                                      , typename Filter<TYPES,_P_>::List
-                                                                      >::Next      
-                                                                      List; };
-    
-    
-    /** append lists-of-types */
-    template<class TY1, class TY2>
-    struct Append                          { typedef Node<TY1, typename Append<TY2,NullType>::List>  List; };
-    
-    template< class TY, class TYPES
-            , class TAIL
-            >
-    struct Append<Node<TY,TYPES>, TAIL>    { typedef Node<TY,  typename Append<TYPES,  TAIL>::List>  List; };
-    
-    template<class TY, class TYPES>
-    struct Append<NullType, Node<TY,TYPES> >   { typedef Node<TY,TYPES>   List; };
-    
-    template<class TY, class TYPES>
-    struct Append<Node<TY,TYPES>, NullType>    { typedef Node<TY,TYPES>   List; };
-    
-    template<class TY1>
-    struct Append<TY1,NullType>            { typedef Node<TY1,NullType>   List; };
-    
-    template<class TY2>
-    struct Append<NullType,TY2>            { typedef Node<TY2,NullType>   List; };
-    
-    template<>
-    struct Append<NullType,NullType>       { typedef NullType             List; };
-    
-    
-    
-    
-    /** access the last list element */
-    template<class TYPES>
-    struct SplitLast;
-    
-    template<>
-    struct SplitLast<NullType>             { typedef NullType Type;
-                                             typedef NullType List; };
-    template<class TY>
-    struct SplitLast<Node<TY,NullType> >   { typedef TY       Type;
-                                             typedef NullType List; };
-    
-    template<class TY, class TYPES>
-    struct SplitLast<Node<TY,TYPES> >      { typedef typename SplitLast<TYPES>::Type Type;
-                                             typedef typename Append< TY,
-                                                                      typename SplitLast<TYPES>::List
-                                                                    >::List 
-                                                                    List; };
-    
-    
-    
-    /** 
-     * splice a typelist like an overlay
-     * into an base typelist, starting at given index.
-     * @return either the combined (spliced) List, or
-     *         the Front/Back part before or after the Overlay
-     * @note using a NullType as OVERLAY allows to extract
-     *         an arbitrary Front/Back part of the list
-     */
-    template<class BASE, class OVERLAY, uint i=0>
-    struct Splice;
-    
-    template<class B, class BS,
-             class OVERLAY, uint i>
-    struct Splice<Node<B,BS>, OVERLAY, i>  { typedef Node<B, typename Splice<BS, OVERLAY, i-1>::List>  List;
-                                             typedef Node<B, typename Splice<BS, OVERLAY, i-1>::Front> Front;
-                                             typedef         typename Splice<BS, OVERLAY, i-1>::Back   Back; };
-    
-    template<class B, class BS,
-             class O, class OS >
-    struct Splice<Node<B,BS>,Node<O,OS>,0> { typedef Node<O, typename Splice<BS,OS, 0>::List>          List;
-                                             typedef NullType                                          Front;
-                                             typedef         typename Splice<BS,OS, 0>::Back           Back; };
-    
-    template<class B, class BS>
-    struct Splice<Node<B,BS>, NullType, 0> { typedef Node<B, BS> List;
-                                             typedef NullType    Front;
-                                             typedef Node<B, BS> Back; };
-    
-    template<class XX, uint i>
-    struct Splice<NullType, XX, i>         { typedef NullType    List;
-                                             typedef NullType    Front;
-                                             typedef NullType    Back; };
-    
-    
-    
-    
     /**
-     * Allows to access various parts of a given typelist:
-     * Start and End, Prefix and Tail..
+     * Metafunction to check if a specific type is contained
+     * in a given typelist. Only exact match is detected.
      */
-    template<class TYPES>
-    struct Dissect;
-    
-    template<class T, class TYPES>
-    struct Dissect<Node<T,TYPES> >
+    template<typename TY, typename TYPES>
+    struct IsInList
       {
-        typedef Node<T,TYPES>                  List;  ///< the complete list
-        typedef T                              Head;  ///< first element
-        typedef Node<T,NullType>               First; ///< a list containing the first element
-        typedef TYPES                          Tail;  ///< remainder of the list starting with the second elm.
-        typedef typename SplitLast<List>::List Prefix;///< all of the list, up to but extcluding the last element
-        typedef typename SplitLast<List>::Type End;   ///< the last element
-        typedef Node<End,NullType>             Last;  ///< a list containing the last element
+        enum{ value = false };
       };
     
-    template<>
-    struct Dissect<NullType>
+    template<typename TY, typename TYPES>
+    struct IsInList<TY, Node<TY,TYPES> >
       {
-        typedef NullType                       List;
-        typedef NullType                       Head;
-        typedef NullType                       First;
-        typedef NullType                       Tail;
-        typedef NullType                       Prefix;
-        typedef NullType                       End;
-        typedef NullType                       Last;
+        enum{ value = true };
       };
     
-    
-    
-    
-    /** 
-     * prefix each of the elements,
-     * yielding a list-of lists-of-types
-     */
-    template<class T, class TY>
-    struct PrefixAll                       { typedef Node< typename Append<T,TY>::List, NullType>  List; };
-    
-    template<class T>
-    struct PrefixAll<T, NullType>          { typedef NullType  List; };
-    
-    template<class T>
-    struct PrefixAll<T, NodeNull>          { typedef Node< typename Append<T,NodeNull>::List, NullType>  List; };
-    
-    template< class T
-            , class TY, class TYPES
-            >
-    struct PrefixAll<T, Node<TY,TYPES> >   { typedef Node< typename Append<T,TY>::List
-                                                         , typename PrefixAll<T,TYPES>::List
-                                                         >     List; };
-    
-    
-    
-    
-    template<class TY1,class TY2>
-    struct Distribute                      { typedef typename PrefixAll<TY1,TY2>::List  List; };
-    
-    template<class TY>
-    struct Distribute<NullType,TY>         { typedef NullType List; };
-    
-    template< class TY, class TYPES
-            , class TAIL
-            >
-    struct Distribute<Node<TY,TYPES>,TAIL> { typedef typename Append< typename PrefixAll<TY,TAIL>::List
-                                                                    , typename Distribute<TYPES,TAIL>::List
-                                                                    >::List                    
-                                                                    List; };
-    
-    
-    
-    /** use a permutation generator
-     *  for creating a list of all possible combinations
-     */
-    template< class X
-            , template<class> class _PERMU_>
-    struct Combine                         { typedef typename Distribute< typename _PERMU_<X>::List
-                                                                        , Node<NullType,NullType>
-                                                                        >::List  List; };
-    template< template<class> class _PERMU_>
-    struct Combine<NullType, _PERMU_ >     { typedef Node<NullType,NullType>     List; };
-    
-    template< class TY, class TYPES
-            , template<class> class _PERMU_>
-    struct Combine<Node<TY,TYPES>,_PERMU_> { typedef typename Distribute< typename _PERMU_<TY>::List
-                                                                        , typename Combine<TYPES,_PERMU_>::List
-                                                                        >::List  List; };
-    
-    /** permutation generator for the Combine metafunction,
-     *  yielding an "on" and "off" case
-     */
-    template<class F>
-    struct FlagOnOff
-      { 
-        typedef Node<F, Node<NullType,NullType> >  List;
-      };
-    
-    
-    /** generate all possible on-off combinations of the given flags */
-    template<class FLAGS>
-    struct CombineFlags
-      { 
-        typedef typename Combine<FLAGS, FlagOnOff>::List  List; 
+    template<typename TY, typename XX, typename TYPES>
+    struct IsInList<TY, Node<XX,TYPES> >
+      {
+        enum{ value = IsInList<TY,TYPES>::value };
       };
     
+    /** convenience shortcut: query function */
+    template<typename TY, typename TYPES>
+    bool
+    isInList()
+    {
+      return IsInList<TY,TYPES>::value;
+    }
     
     
     

src/lib/meta/typelist.hpp

@@ -40,15 +40,25 @@ This code is heavily inspired by
 
 /** @file typelist.hpp
  ** A template metaprogramming technique for manipulating collections of types.
- ** Effectively this is a taylored and simplified version of what can be found in the Loki library.
+ ** Effectively this is a tailored and simplified version of what can be found in the Loki library.
  ** We use it in other generic library-style code to generate repetitive code. If you tend to find
  ** template metaprogramming (or functional programming in general) offending, please ignore the 
- ** technical details and just consider the benefit of such an simplification for the user code.
- **
- ** Interface for using this facility is the template Types(.....) for up to 20 Type parameters
+ ** technical details and just consider the benefit of such an simplification for the client code.
  **
+ ** Interface for using this facility is the template Types(.....) for up to 20 Type parameters.
+ ** To start typelist processing, other templates typically pick up the Types<...>::List type.
+ ** This allows for LISP-style list processing, with a pattern match on either Node<TY,TYPES>
+ ** or NullType to terminate recursion. In C++ template metaprogramming, "pattern match"
+ ** is done by partial template specialisations (the compiler will pick up and thus
+ ** match the template parameters). A typedef acts like a declaration in normal
+ ** programming. Because such a "declaration" can't be changed after the fact,
+ ** effectively this is a flavour of functional programming. Just the
+ ** "execution environment" is the compiler, during compilation.
+ ** 
  ** @see lumiera::visitor::Applicable usage example
- ** @see typelisttest.cpp
+ ** @see control::CommandSignature more elaborate usage example (dissecting a functor signature)
+ ** @see TypeList_test
+ ** @see TypeListManip_test
  ** 
  */
 

src/lib/meta/typeseq-util.hpp

@@ -45,7 +45,7 @@
 #define LUMIERA_META_TYPESEQ_UTIL_H
 
 #include "lib/meta/typelist.hpp"
-#include "lib/meta/typelist-util.hpp"
+#include "lib/meta/typelist-manip.hpp"
 #include "lib/meta/util.hpp"
 
 

src/lib/meta/util.hpp

@@ -38,6 +38,24 @@ namespace lumiera {
     
   namespace typelist {
     
+    /** Compile-time Type equality:
+     *  Simple Trait template to pick up types considered
+     *  \em identical by the compiler.
+     * @warning identical, not sub-type!
+     */
+    template<typename T1, typename T2>
+    struct is_sameType
+      {
+        static const bool value = false;
+      };
+    
+    template<typename T>
+    struct is_sameType<T,T>
+      {
+        static const bool value = true;
+      };
+    
+    
     /** semi-automatic detection if an instantiation is possible.
      *  Requires help by the template to be tested, which needs to define
      *  a typedef member \c is_defined. The embedded metafunction Test can be used

src/lib/opaque-holder.hpp

@@ -219,7 +219,7 @@ namespace lib {
           char content_[siz];
           void* ptr() { return &content_; }
           
-          virtual ~Buffer() {}
+          virtual ~Buffer() {}           ///< this is an ABC with VTable
           virtual bool  isValid()  const =0;
           virtual bool  empty()    const =0;
           virtual BaseP getBase()  const =0;

src/lib/p.hpp

@@ -79,51 +79,51 @@ namespace lumiera {
                                  P ( )                      : BASE() {}
       template<class Y> explicit P (Y* p)                   : BASE(p) {}
       template<class Y, class D> P (Y* p, D d)              : BASE(p,d){}
-
+      
                                  P (P const& r)             : BASE(r)  {}
       template<class Y>          P (shared_ptr<Y> const& r) : BASE(r)  {}
       template<class Y> explicit P (weak_ptr<Y> const& wr)  : BASE(wr) {}
-      template<class Y> explicit P (std::auto_ptr<Y> & ar)  : BASE(ar) {}      
-    
+      template<class Y> explicit P (std::auto_ptr<Y> & ar)  : BASE(ar) {}
       
-      P& operator= (P const& r)                               { BASE::operator= (r);  return *this; }  
+      
+      P& operator= (P const& r)                               { BASE::operator= (r);  return *this; }
       template<class Y> P& operator=(shared_ptr<Y> const& sr) { BASE::operator= (sr); return *this; }
       template<class Y> P& operator=(std::auto_ptr<Y> & ar)   { BASE::operator= (ar); return *this; }
-
+      
       TAR* get() const        { return dynamic_cast<TAR*> (BASE::get()); }
       TAR& operator*() const  { return *get(); }
       TAR* operator->() const { return get();  }
-
+      
       void swap(P& b)         { BASE::swap (b);}
       
-    
+      
     private: /* === friend operators injected into enclosing namespace for ADL === */
       template<typename _O_>
       friend inline bool
       operator== (P const& p, P<_O_> const& q) { return (p && q)? (*p == *q) : (!p && !q); }
-  
+      
       template<typename _O_>
       friend inline bool
       operator!= (P const& p, P<_O_> const& q) { return (p && q)? (*p != *q) : !(!p && !q); }
-  
+      
       template<typename _O_>
       friend inline bool
       operator<  (P const& p, P<_O_> const& q) { REQUIRE (p && q); return *p < *q; }   ///< @note deliberately not allowing comparison on NIL ////TICKET #307  : problem with equality test in associative containers, where equal(a,b) := !(a < b) && !(b < a)
-  
+      
       template<typename _O_>
       friend inline bool
       operator>  (P const& p, P<_O_> const& q) { REQUIRE (p && q); return *q < *p; }
-
+      
       template<typename _O_>
       friend inline bool
       operator<= (P const& p, P<_O_> const& q) { REQUIRE (p && q); return *p <= *q;}
-
+      
       template<typename _O_>
       friend inline bool
       operator>= (P const& p, P<_O_> const& q) { REQUIRE (p && q); return *p >= *q;}
       
     };
-   
-    
+  
+  
 } // namespace lumiera
 #endif

src/lib/polymorphic-value.hpp

@@ -389,21 +389,21 @@ namespace lib {
         }
       
       template<class IMP, typename A1>
-      PolymorphicValue (IMP*, A1& a1)
+      PolymorphicValue (IMP*, A1 a1)
         {
           REQUIRE (siz >= sizeof(IMP));
           new(&buf_) IMP (a1);
         }
       
       template<class IMP, typename A1, typename A2>
-      PolymorphicValue (IMP*, A1& a1, A2& a2)
+      PolymorphicValue (IMP*, A1 a1, A2 a2)
         {
           REQUIRE (siz >= sizeof(IMP));
           new(&buf_) IMP (a1,a2);
         }
       
       template<class IMP, typename A1, typename A2, typename A3>
-      PolymorphicValue (IMP*, A1& a1, A2& a2, A3& a3)
+      PolymorphicValue (IMP*, A1 a1, A2 a2, A3 a3)
         {
           REQUIRE (siz >= sizeof(IMP));
           new(&buf_) IMP (a1,a2,a3);
@@ -444,13 +444,13 @@ namespace lib {
           Adapter() : IMP() { }
           
           template<typename A1>
-          Adapter (A1& a1) : IMP(a1) { }
+          Adapter (A1 a1) : IMP(a1) { }
           
           template<typename A1, typename A2>
-          Adapter (A1& a1, A2& a2) : IMP(a1,a2) { }
+          Adapter (A1 a1, A2 a2) : IMP(a1,a2) { }
           
           template<typename A1, typename A2, typename A3>
-          Adapter (A1& a1, A2& a2, A3& a3) : IMP(a1,a2,a3) { }
+          Adapter (A1 a1, A2 a2, A3 a3) : IMP(a1,a2,a3) { }
           
           /* using default copy and assignment */
         };
@@ -505,31 +505,31 @@ namespace lib {
       static PolymorphicValue
       build ()
         {
-          Adapter<IMP>* type_to_build_in_buffer;
+          Adapter<IMP>* type_to_build_in_buffer(0);
           return PolymorphicValue (type_to_build_in_buffer);
         }
       
       template<class IMP, typename A1>
       static PolymorphicValue
-      build (A1& a1)
+      build (A1 a1)
         {
-          Adapter<IMP>* type_to_build_in_buffer;
+          Adapter<IMP>* type_to_build_in_buffer(0);
           return PolymorphicValue (type_to_build_in_buffer, a1);
         }
       
       template<class IMP, typename A1, typename A2>
       static PolymorphicValue
-      build (A1& a1, A2& a2)
+      build (A1 a1, A2 a2)
         {
-          Adapter<IMP>* type_to_build_in_buffer;
+          Adapter<IMP>* type_to_build_in_buffer(0);
           return PolymorphicValue (type_to_build_in_buffer, a1,a2);
         }
       
       template<class IMP, typename A1, typename A2, typename A3>
       static PolymorphicValue
-      build (A1& a1, A2& a2, A3& a3)
+      build (A1 a1, A2 a2, A3 a3)
         {
-          Adapter<IMP>* type_to_build_in_buffer;
+          Adapter<IMP>* type_to_build_in_buffer(0);
           return PolymorphicValue (type_to_build_in_buffer, a1,a2,a3);
         }
       

src/lib/result.hpp

@@ -24,9 +24,9 @@
 /** @file result.hpp
  ** Intermediary value object to represent the result of an operation.
  ** This operation might have produced a value result or failed with an exception.
- ** Typically, the Result token used \em inline -- immediately either invoking one
- ** of the member function or employing the built-in result type conversion. It
- ** will be copyable iff the result value is copyable. There is an implicit
+ ** Typically, the Result token is used \em inline -- immediately either invoking
+ ** one of the member function or employing the built-in result type conversion.
+ ** It will be copyable iff the result value is copyable. There is an implicit
  ** valid or failure state, which can be tested. Any attempt to get the value
  ** of an invalid result token will cause in an exception to be thrown.
  ** 
@@ -58,11 +58,10 @@ namespace lib {
   
   
   /**
-   * Result value and status of some operation.
-   * It can be created for passing a result produced
-   * by the operation, or the failure to do so. The value
-   * can be retrieved by implicit or explicit conversion.
-   * @throws on any attempt to access the value in case of failure
+   * Optional Result value or status of some operation.
+   * It can be created for passing a result produced by the operation, or the 
+   * failure to do so. The value can be retrieved by implicit or explicit conversion.
+   * @throw error::State on any attempt to access the value in case of failure
    * @warning this class has a lot of implicit conversions;
    *          care should be taken when defining functions
    *          to take Result instances as parameter....

src/lib/scoped-holder-transfer.hpp

@@ -105,8 +105,8 @@ namespace lib {
       pointer       address(reference r)            const { return par_.address(r); }
       const_pointer address(const_reference cr)     const { return par_.address(cr); }
       pointer       allocate(size_type n, const void *p=0){ return par_.allocate(n,p); }
-      void          deallocate(pointer p, size_type n)    { return par_.deallocate(p,n); }
-      void          destroy(pointer p)                    { return par_.destroy(p); }
+      void          deallocate(pointer p, size_type n)    {        par_.deallocate(p,n); }
+      void          destroy(pointer p)                    {        par_.destroy(p); }
       
       
       void 

src/lib/scoped-holder.hpp

@@ -173,11 +173,21 @@ namespace lib {
           return *obj;
         }
       
+      TY& 
+      create (TY const& o)    ///< place new content object using copy ctor
+        {
+          ASSERT (!created_);
+          TY * obj = new(content_) TY(o);
+          ++created_;
+          return *obj;
+        }
+      
       void
       clear ()
         {
           if (created_)
             get()->~TY();
+          created_ = false;
         }
       
       

src/lib/simple-allocator.hpp

@@ -0,0 +1,302 @@
+/*
+  SIMPLE-ALLOCATOR.hpp  -  frontend for plain explicit allocations 
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+/** @file simple-allocator.hpp
+ ** Frontend and marker interface for allocating small objects explicitly.
+ ** Contrary to the TypedAllocationManager, the SimpleAllocator doesn't provide any
+ ** ref-counting or tracking facilities, nor does he support bulk de-allocation.
+ ** Each object needs to be allocated and released by explicit call.
+ ** The advantage over using std::allocator  directly is the shortcut for (placement) construction,
+ ** and -- of course -- the ability to exchange the memory model at one central location.
+ ** 
+ ** SimpleAllocator instances will be defined for a specific collection of types; for each of those
+ ** types, there will be an embedded dedicated custom allocator (currently as of 9/2011, just
+ ** implemented as std::allocator<TY>). Objects of these preconfigured types can be constructed
+ ** and destroyed through this allocator instance. Each call needs to be done explicitly, with
+ ** the precise, concrete type to be created or destroyed. This is especially important for
+ ** the releasing of objects: there is \em no support for any kind of virtual destruction.
+ ** 
+ ** @see engine::BufferMetadata
+ ** @see TypedAllocationManager (another more elaborate custom allocation scheme)
+ **
+ */
+
+
+
+#ifndef LIB_SIMPLE_ALLOCATOR_H
+#define LIB_SIMPLE_ALLOCATOR_H
+
+//#include "pre.hpp"
+#include "lib/error.hpp"
+#include "lib/meta/generator.hpp"
+#include "lib/meta/typelist-util.hpp"
+#include "lib/format.hpp"
+#include "lib/typed-counter.hpp"
+#include "include/logging.h"
+
+#include <boost/static_assert.hpp>
+#include <memory>
+
+
+
+namespace lib {
+  
+  using lumiera::typelist::Types;
+  using lumiera::typelist::IsInList;
+  using lumiera::typelist::InstantiateForEach;
+  
+  
+  
+  /* === Policies for simple custom allocator  === */
+  
+  /**
+   * Policy: use just plain heap allocations
+   */
+  template<typename TY>
+  class CustomAllocator
+    : public std::allocator<TY>
+    { };
+  
+  
+  /**
+   * Policy: maintain explicit per type instance count
+   * @note this imposes additional locking
+   */
+  struct UseInstantiationCounting
+    {
+      template<class XX>
+      size_t
+      allocationCount()  const
+        {
+          return allocCnt_.get<XX>();
+        }
+      
+      template<class XX>
+      void
+      incrementCount()
+        {
+          allocCnt_.inc<XX>();
+        }
+      
+      template<class XX>
+      void
+      decrementCount()
+        {
+          allocCnt_.dec<XX>();
+        }
+      
+    private:
+      lib::TypedCounter allocCnt_;
+    };
+  
+  /**
+   * Policy: no additional instantiation accounting
+   */
+  struct NoInstantiationCount
+    {
+      template<class XX>  size_t allocationCount()  const { return 0; }
+      template<class XX>  void    incrementCount()  { /* NOP */ }
+      template<class XX>  void    decrementCount()  { /* NOP */ }
+    };
+    
+  
+  
+  
+  
+  
+  /* === Allocator frontend === */
+  
+  /**
+   * Frontend for explicit allocations, using a custom allocator.
+   * This template is to be instantiated for the collection of types
+   * later to be allocated through this custom memory manager/model.
+   * It provides convenience shortcuts for placement-construction
+   * and releasing of target objects.
+   * 
+   * @todo currently (as of 8/09) the low-level pooled allocator
+   *       isn't implemented; instead we do just heap allocations.
+   *       ////////////////////////////////////////////////////////////////////////////////////////////Ticket #835
+   */
+  template<typename TYPES
+          ,class COUNTER = NoInstantiationCount       ///< Policy: support instance accounting?
+          >
+  class SimpleAllocator
+    : InstantiateForEach< typename TYPES::List     // for each of those types...
+                        , CustomAllocator         //  ...mix in the custom allocator
+                        >
+    , COUNTER
+    {
+      
+      /** forward plain memory allocation */
+      template<class XX>
+      XX *
+      allocateSlot ()
+        {
+          TRACE (memory, "allocate %s", util::tyStr<XX>().c_str());
+          XX * newStorage = CustomAllocator<XX>::allocate (1);
+          COUNTER::template incrementCount<XX>();
+          return newStorage;
+        }
+      
+      template<class XX>
+      void
+      releaseSlot (XX* entry)
+        {
+          TRACE (memory, "release %s", util::tyStr<XX>().c_str());
+          CustomAllocator<XX>::deallocate (entry, 1);
+          COUNTER::template decrementCount<XX>();
+        }
+      
+      
+      template<class XX>
+      void
+      ___assertSupportedType()
+        {
+          typedef typename TYPES::List PreconfiguredTypes;
+          typedef IsInList<XX, PreconfiguredTypes> IsSupportedType;
+          
+          BOOST_STATIC_ASSERT (IsSupportedType::value);
+        }
+      
+      
+      
+      
+      public: /* ==== build objects with managed allocation ==== */
+      
+#define _EXCEPTION_SAFE_INVOKE(_CTOR_)                      \
+                                                             \
+          ___assertSupportedType<XX>();                       \
+          XX* storage = allocateSlot<XX>();                    \
+          try                                                   \
+            {                                                    \
+              return (new(storage) _CTOR_ );                     \
+            }                                                    \
+          catch(...)                                             \
+            {                                                    \
+              releaseSlot<XX>(storage);                          \
+              throw;                                             \
+            }
+      
+      template< class XX>
+      XX*                                                                              //_____________________
+      create ()                                                                       ///< invoke default ctor
+        {
+          _EXCEPTION_SAFE_INVOKE ( XX() )
+        }
+      
+      
+      template< class XX, typename P1>
+      XX*                                                                              //___________________
+      create (P1& p1)                                                                 ///< invoke 1-arg ctor
+        {
+          _EXCEPTION_SAFE_INVOKE ( XX (p1) )
+        }
+      
+      
+      template< class XX
+              , typename P1
+              , typename P2
+              >
+      XX*                                                                              //___________________
+      create (P1& p1, P2& p2)                                                         ///< invoke 2-arg ctor
+        {
+          _EXCEPTION_SAFE_INVOKE ( XX (p1,p2) )
+        }
+      
+      
+      template< class XX
+              , typename P1
+              , typename P2
+              , typename P3
+              >
+      XX*                                                                              //___________________
+      create (P1& p1, P2& p2, P3& p3)                                                 ///< invoke 3-arg ctor
+        {
+          _EXCEPTION_SAFE_INVOKE ( XX (p1,p2,p3) )
+        }
+      
+      
+      template< class XX
+              , typename P1
+              , typename P2
+              , typename P3
+              , typename P4
+              >
+      XX*                                                                              //___________________
+      create (P1& p1, P2& p2, P3& p3, P4& p4)                                         ///< invoke 4-arg ctor
+        {
+          _EXCEPTION_SAFE_INVOKE ( XX (p1,p2,p3,p4) )
+        }
+      
+      
+      template< class XX
+              , typename P1
+              , typename P2
+              , typename P3
+              , typename P4
+              , typename P5
+              >
+      XX*                                                                              //___________________
+      create (P1& p1, P2& p2, P3& p3, P4& p4, P5& p5)                                 ///< invoke 5-arg ctor
+        {
+          _EXCEPTION_SAFE_INVOKE ( XX (p1,p2,p3,p4,p5) )
+        }
+      
+#undef _EXCEPTION_SAFE_INVOKE
+      
+      
+      
+      template<class XX>
+      void
+      destroy (XX* entry)
+        {
+          if (!entry) return;
+          ___assertSupportedType<XX>();
+          try
+            {
+              entry->~XX();
+            }
+          catch(...)
+            {
+              lumiera_err errorID = lumiera_error();
+              WARN (common_dbg, "dtor of %s failed: %s", util::tyStr(entry).c_str()
+                                                       , errorID );
+            }
+          releaseSlot<XX> (entry);
+        }
+      
+      
+      /** diagnostics */
+      template<class XX>
+      size_t
+      numSlots()  const
+        {
+          return COUNTER::template allocationCount<XX>();
+        }
+    };
+  
+  
+  
+  
+} // namespace lib
+#endif

src/lib/streamtype.hpp

@@ -110,7 +110,12 @@ namespace lumiera {
       
       class TypeTag ; 
       
-      /** placeholder definition for the contents of a data buffer */
+      /** 
+       * placeholder type for the contents of a data buffer.
+       * The actual buffer will always be provided by a 
+       * library implementation; throughout the engine,
+       * it's just hidden behind a DataBuffer pointer.
+       */
       struct DataBuffer { };
       
       

src/lib/symbol-impl.cpp

@@ -50,6 +50,9 @@ using boost::hash_combine;
 
 namespace lib {
   
+  const size_t STRING_MAX_RELEVANT = 1000;
+  
+  
   /** equality on Symbol values is defined
    *  based on the content, not the address. */
   bool

src/lib/symbol.hpp

@@ -112,7 +112,7 @@ namespace lib {
   /** safety guard: maximum number of chars to process.
    *  For comparisons, hash calculations etc., when dealing
    *  with raw char ptrs (typically literal values) */
-  const size_t STRING_MAX_RELEVANT = 1000;
+  extern const size_t STRING_MAX_RELEVANT;
   
   
   /* ===== to be picked up by ADL ===== */

src/lib/test/suite.cpp

@@ -37,6 +37,8 @@
 #include <memory>
 #include <vector>
 #include <map>
+#include <cstdlib>
+#include <ctime>
 
 
 namespace test {
@@ -141,6 +143,9 @@ namespace test {
     REQUIRE( !isnil(groupID) );
     TRACE(test, "Test-Suite( groupID=%s )\n", groupID.c_str () );
     
+    // Seed random number generator
+    std::srand (std::time (NULL));
+    
     if (!testcases.getGroup(groupID))
       throw lumiera::error::Invalid ("empty testsuite");
   }

src/lib/time/control-impl.hpp

@@ -0,0 +1,203 @@
+/*
+  CONTROL-IMPL.hpp  -  time::control implementation building blocks  
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+/** @file control-impl.hpp
+ ** Implementation building blocks for time modification and propagation.
+ ** The time::Control element allows to impose modifications to a connected
+ ** time value entity and at the same time publish the changes to registered
+ ** listeners. Due to the various flavours of actual time value entities, this
+ ** is a complex undertaking, which is implemented here based on policies and
+ ** template metaprogramming. This header/include defines two building blocks:
+ ** - the actual Mutator to apply the changes to the target entity
+ ** - a Propagator to register listeners and forward the changes.
+ ** 
+ ** \par implementation technique
+ ** 
+ ** The Mutator uses functor objects to encapsulate the actual modification
+ ** operations. When attaching to a target time entity to be manipulated, these
+ ** functor objects will be configured by binding them to the appropriate
+ ** implementation function. And picking this actual implementation is done
+ ** through a time::mutation::Policy element, using the concrete time entity
+ ** types as template parameter. Thus, the actual implementation to be used
+ ** is determined by the compiler, through the template specialisations
+ ** contained in control-policy.hpp
+ ** 
+ ** @note the header control-policy.hpp with the template specialisations
+ **       is included way down, after the class definitions. This is done
+ **       so for sake of readability
+ ** 
+ ** @see TimeControl_test
+ **
+ */
+
+#ifndef LIB_TIME_CONTROL_IMPL_H
+#define LIB_TIME_CONTROL_IMPL_H
+
+#include "lib/error.hpp"
+#include "lib/time/mutation.hpp"
+#include "lib/time/timevalue.hpp"
+
+#include <vector>
+
+
+namespace lib {
+namespace time {
+namespace mutation {
+  
+  
+  
+  
+  /**
+   * Implementation building block: impose changes to a Time element.
+   * The purpose of the Mutator is to attach a target time entity,
+   * which then will be subject to any received value changes,
+   * offsets and grid nudging. The actual attachment is to be
+   * performed in a subclass, by using the Mutation interface.
+   * When attaching to a target, the Mutator will be outfitted 
+   * with a set of suitable functors, incorporating the specific
+   * behaviour for the concrete combination of input changes
+   * ("source values") and target object type. This works by
+   * binding to the appropriate implementation functionality,
+   * guided by a templated policy class. After installing
+   * these functors, these decisions remains opaque and
+   * encapsulated within the functor objects, so the
+   * mutator object doesn't need to carry this 
+   * type information on the interface
+   */
+  template<class TI>
+  class Mutator
+    : public Mutation
+    {
+      typedef function<TI(TI const&)>     ValueSetter;
+      typedef function<TI(Offset const&)> Ofsetter;
+      typedef function<TI(int)>           Nudger;
+      
+    protected:
+      mutable ValueSetter setVal_;
+      mutable Ofsetter    offset_;
+      mutable Nudger      nudge_;
+      
+      void
+      ensure_isArmed()  const
+        {
+          if (!setVal_)
+            throw error::State("feeding time/value change "
+                               "while not (yet) connected to any target to change"
+                              ,error::LUMIERA_ERROR_UNCONNECTED);
+        }
+      
+      
+      template<class TAR>
+      void bind_to (TAR& target)  const;
+      
+      void unbind();
+      
+      // using default construction and copy
+    };
+  
+  
+  
+  /**
+   * Implementation building block: propagate changes to listeners.
+   * The Propagator manages a set of callback signals, allowing to
+   * propagate notifications for changed Time values.
+   * 
+   * There are no specific requirements on the acceptable listeners,
+   * besides exposing a function-call operator to feed the changed
+   * time value to. Both Mutator and Propagator employ one primary
+   * template parameter, which is the type of the time values
+   * to be fed in and propagated. 
+   */
+  template<class TI>
+  class Propagator
+    {
+      typedef function<void(TI const&)> ChangeSignal;
+      typedef std::vector<ChangeSignal> ListenerList;
+      
+      ListenerList listeners_;
+      
+    public:
+      /** install notification receiver */
+      template<class SIG>
+      void
+      attach (SIG const& toNotify)
+        {
+          ChangeSignal newListener (ref(toNotify));
+          listeners_.push_back (newListener);
+        }
+      
+      /** disconnect any observers */
+      void
+      disconnnect()
+        {
+          listeners_.clear();
+        }
+      
+      /** publish a change */
+      TI
+      operator() (TI const& changedVal)  const
+        {
+          typedef typename ListenerList::const_iterator Iter;
+          Iter p = listeners_.begin();
+          Iter e = listeners_.end();
+          
+          for ( ; p!=e; ++p )
+            (*p) (changedVal);
+          return changedVal;
+        }
+      
+      // using default construction and copy
+    };
+  
+}}} // lib::time::mutation
+
+
+/* ===== Definition of actual operations ===== */
+#include "lib/time/control-policy.hpp"
+
+
+
+
+template<class TI>
+template<class TAR>
+void
+lib::time::mutation::Mutator<TI>::bind_to (TAR& target)  const
+{
+  using lib::time::mutation::Policy;
+  
+  setVal_ = Policy<TI,TI,    TAR>::buildChangeHandler (target);
+  offset_ = Policy<TI,Offset,TAR>::buildChangeHandler (target);
+  nudge_  = Policy<TI,int,   TAR>::buildChangeHandler (target);
+}
+
+template<class TI>
+void
+lib::time::mutation::Mutator<TI>::unbind()
+{
+  setVal_ = ValueSetter();
+  offset_ = Ofsetter();
+  nudge_  = Nudger();
+}
+
+
+#endif

src/lib/time/control-policy.hpp

@@ -0,0 +1,400 @@
+/*
+  CONTROL-POLICY.hpp  -  detail definition of actual time changing functionality  
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+/** @file control-policy.hpp
+ ** Definition of special cases when imposing a change onto concrete time values.
+ ** The time::Control element allows to impose modifications to a connected
+ ** time value entity and at the same time publish the changes to registered
+ ** listeners. Due to the various flavours of actual time value entities, this
+ ** is a complex undertaking, which is implemented here based on policies and
+ ** template metaprogramming.
+ ** 
+ ** The header control-impl.hpp defines the building blocks for time::Control
+ ** and then includes this header here to get the concrete template specialisations
+ ** for time::mutation::Policy. This policy class is templated by time entity types
+ ** - for \c TI, the \em nominal value type used on the time::Control interface
+ ** - for \c SRC, the actual type of values to impose as \em change
+ ** - for \c TAR, the target time value's type, receiving those changes.
+ ** 
+ ** \par mutating a time value entity
+ ** 
+ ** Actually imposing a change to the attached time value entity involves several
+ ** steps. Each of these steps might be adapted specifically, in accordance to
+ ** the concrete time value types involved.
+ ** - TimeValue, Time
+ ** - Offset
+ ** - Duration
+ ** - TimeSpan
+ ** - QuTime (grid aligned time value)
+ ** - QuTimeSpan (planned as of 6/2011)
+ ** 
+ ** Moreover, the combination of types needs to be taken into account. For example,
+ ** it doesn't make sense to apply a Duration value as change to a TimeValue, which
+ ** has no duration (temporal extension). While a TimeSpan might receive a Duration
+ ** change, but behaves differently when imposing a Time to manipulate the starting
+ ** point of the time interval given by the TimeSpan.
+ ** 
+ ** Incoming changes might be of any of the aforementioned types, and in addition,
+ ** we might receive \em nudging, which means to increment or decrement the target
+ ** time value in discrete steps. After maybe adapting these incoming change values,
+ ** they may be actually \em imposed to the target. In all cases, this is delegated
+ ** to the time::Mutation base class, which is declared fried to TimeValue and thus
+ ** has the exceptional ability to manipulate time values, which otherwise are defined
+ ** to be immutable. Additionally, these protected functions in the time::Mutation
+ ** baseclass also know how to handle \em nudge values, either by using the native
+ ** (embedded) time grid of a quantised time value, or by falling back to a standard
+ ** nudging grid, defined in the session context (TODO as of 6/2011).                     //////////////////////TICKET #810
+ ** 
+ ** After (maybe) imposing a change to the target, the change \em notification value
+ ** needs to be built. This is the time value entity to be forwarded to registered
+ ** listeners. This notification value has to be given as the type \c TI, in accordance
+ ** to the \c time::Control<TI> frontend definition used in the concrete usage situation.
+ ** As this type \c TI might be different to the actual target type, and again different
+ ** to the type of the change handed in, in some cases this involves a second conversion
+ ** step, to represent the current state of the target \c TAR in terms of the interface
+ ** type \c TI.
+ ** 
+ ** \par changing quantised (grid aligned) time entities
+ ** 
+ ** The time::Control element includes the capability to handle grid aligned time values,
+ ** both as target and as change/notification value. This ability is compiled in conditionally,
+ ** as including mutation.hpp causes several additional includes, which isn't desirable when
+ ** it comes just to changing plain time values. Thus, to get these additional specialisations,
+ ** the LIB_TIME_TIMEQUQNT_H header guard needs to be defined, which happens automatically
+ ** if lib/time/mutation.hpp is included prior to lib/time/control.hpp.
+ ** 
+ ** As a special convention, any \em quantised (grid aligned) types involved in these
+ ** time changes will be \em materialised, whenever a type conversion happens. Generally
+ ** speaking, a quantised time value contains an (opaque) raw time value, plus a reference
+ ** to a time grid definition to apply. In this context \em materialising means actually
+ ** to apply this time grid to yield a grid aligned value. Thus, when using a quantised
+ ** value to impose as change (or to receive a change), its grid aligning nature
+ ** becomes effective, by applying the \em current definition of the grid to
+ ** create a fixed (materialised) time value, aligned to that current grid.
+ ** 
+ ** @todo 6/2011 include all the special cases for QuTimeSpan                               ////////////////////TICKET #760
+ ** 
+ ** @see TimeControl_test
+ **
+ */
+
+#ifndef LIB_TIME_CONTROL_POLICY_H
+#define LIB_TIME_CONTROL_POLICY_H
+
+#include "lib/meta/util.hpp"
+#include "lib/time/mutation.hpp"
+#include "lib/time/timevalue.hpp"
+
+#include <boost/utility/enable_if.hpp>
+#include <tr1/functional>
+
+
+namespace lib {
+namespace time {
+  
+namespace mutation {
+  
+  using boost::disable_if;
+  using lumiera::typelist::is_sameType;
+  using std::tr1::placeholders::_1;
+  using std::tr1::function;
+  using std::tr1::bind;
+  using std::tr1::ref;
+  
+  
+  
+  namespace { // metaprogramming helpers to pick a suitable implementation branch...
+    
+    template<class T>
+    inline bool
+    isDuration()
+    {
+      return is_sameType<T,Duration>::value;
+    }
+    
+    template<class T>
+    inline bool
+    isTimeSpan()
+    {
+      return is_sameType<T,TimeSpan>::value;
+    }
+    
+    template<class T>
+    inline T const&
+    maybeMaterialise (T const& non_grid_aligned_TimeValue)
+    {
+      return non_grid_aligned_TimeValue;
+    }
+    
+#ifdef LIB_TIME_TIMEQUQNT_H
+    inline QuTime
+    maybeMaterialise (QuTime const& alignedTime)
+    {
+      PQuant const& grid(alignedTime);
+      return QuTime(grid->materialise(alignedTime), grid);
+    }
+#endif //--quantised-time-support
+  }
+  
+  
+  /** 
+   * Implementation policy: how to build a new
+   * notification value of type \c TI, given a 
+   * target time value entity of type \c TAR
+   */
+  template<class TI, class TAR>
+  struct Builder
+    {
+      static TI
+      buildChangedValue (TAR const& target)
+        {
+          return TI(target);
+        }
+    };
+  template<class TAR>
+  struct Builder<TimeSpan, TAR>
+    {
+      static TimeSpan
+      buildChangedValue (TAR const& target)
+        {
+          return TimeSpan (target, Duration::NIL);
+        }
+    };
+  template<>
+  struct Builder<TimeSpan, Duration>
+    {
+      static TimeSpan
+      buildChangedValue (Duration const& targetDuration)
+        {
+          return TimeSpan (Time::ZERO, targetDuration);
+        }
+    };
+  template<>
+  struct Builder<TimeSpan, TimeSpan>
+    {
+      static TimeSpan
+      buildChangedValue (TimeSpan const& target)
+        {
+          return target;
+        }
+    };
+#ifdef LIB_TIME_TIMEQUQNT_H
+  template<class TAR>
+  struct Builder<QuTime, TAR>
+    {
+      static QuTime
+      buildChangedValue (TAR const& target)
+        {
+          return QuTime (target
+                        ,getDefaultGridFallback()                                                     //////////////////TICKET #810
+                        );
+        }
+    };
+  template<>
+  struct Builder<QuTime, QuTime>
+    {
+      static QuTime
+      buildChangedValue (QuTime const& target)
+        {
+          return target;
+        }
+    };
+#endif //--quantised-time-support
+  
+  
+  
+  
+  /**  
+   * Policy to tie the various detail policies together
+   * for providing actual value change operations.
+   * The standard case uses the (inherited) time::Mutation
+   * base implementation to impose a new value onto the
+   * target entity and then uses the Builder policy to
+   * create a notification value reflecting this change.
+   */
+  template<class TI, class TAR>
+  struct Link
+    : Mutation
+    , Builder<TI,TAR>
+    {
+      
+      template<class SRC>
+      static TI
+      processValueChange (TAR& target, SRC const& change)  ///< standard case: plain value change
+        {
+          imposeChange (target, maybeMaterialise(change));
+          return buildChangedValue (maybeMaterialise(target));
+        }
+      
+      static TI
+      useLengthAsChange (TAR& target, TimeSpan const& change)
+        {
+          return processValueChange(target, change.duration());
+        }
+      
+      static TI
+      mutateLength (TimeSpan& target, Duration const& change)
+        {
+          imposeChange (target.duration(), change);
+          return Builder<TI,TimeSpan>::buildChangedValue(target);
+        }
+      
+      static TimeSpan
+      mutateTimeSpan (TimeSpan& target, TimeSpan const& change)
+        {
+          imposeChange (target.duration(), change.duration());
+          imposeChange (target,change.start());
+          return Builder<TimeSpan,TimeSpan>::buildChangedValue(target);
+        }
+      
+      static TI
+      dontChange (TAR& target)  ///< @note: not touching the target
+        {
+          return buildChangedValue(target);
+        }
+    };
+  
+  
+  
+  /** 
+   * Policy how to impose changes onto a connected target time value entity
+   * This policy will be parametrised with the concrete time entity types
+   * involved in the usage situation of time::Control. The purpose of the
+   * policy is to \em bind a functor object to the concrete implementation
+   * of the value change applicable for this combination of types.
+   * This functor will then be stored within time::Control and
+   * invoked for each actual value change.
+   * @param TI the nominal (interface) type of the change, propagated to listeners
+   * @param SRC the actual type of the change to be imposed
+   * @param TAR the actual type of the target entity to receive the changes
+   * @note typically either SRC is identical to TI, or it is an
+   *       time::Offset, or an int for \em nudging the target 
+   */
+  template<class TI, class SRC, class TAR>
+  struct Policy
+    {
+      static function<TI(SRC const&)>
+      buildChangeHandler (TAR& target)
+        {
+          return bind (Link<TI,TAR>::template processValueChange<SRC>, ref(target), _1 );
+        }
+    };
+  
+  
+  // special treatment of Durations as target------------------------------------
+  
+  namespace {
+    template<class T>
+    struct canMutateDuration
+      {
+        static const bool value = is_sameType<T,Duration>::value
+                             ||   is_sameType<T,Offset>::value
+                             ||   is_sameType<T,int>::value;
+      };
+    
+    template<class T>
+    struct canReceiveDuration
+      {
+        static const bool value = is_sameType<T,Duration>::value
+                             ||   is_sameType<T,TimeSpan>::value;
+      };
+  }
+  
+  
+  /** 
+   * special case: a Duration target value can't be changed by plain time values.
+   * This specialisation is \em not used (\c disable_if ) when the given change (SRC)
+   * is applicable to a Duration in a sensible way. We either define explicit
+   * specialisations (for TimeSpan) or fall back to the default in these cases. 
+   */
+  template<class TI, class SRC>
+  struct Policy<TI,SRC,                  typename disable_if< canMutateDuration<SRC>, 
+                       Duration>::type>
+    {
+      static function<TI(SRC const&)>
+      buildChangeHandler (Duration& target)
+        {
+          return bind (Link<TI,Duration>::dontChange, ref(target) );
+        }
+    };
+  
+  /**
+   * special case: a Duration change value can't be imposed to a plain time value.
+   * In these cases, we even propagate a Duration::ZERO to the listeners.
+   * As above, there are exceptions to this behaviour, where a Duration change
+   * can sensibly be applied. 
+   */
+  template<class TAR>
+  struct Policy<Duration,                typename disable_if< canReceiveDuration<TAR>, 
+                   Duration>::type, TAR>
+    {
+      static function<Duration(Duration const&)>
+      buildChangeHandler (TAR&)
+        {
+          return bind ( ignore_change_and_return_Zero );
+        }
+      
+      static Duration
+      ignore_change_and_return_Zero()
+        {
+          return Duration::NIL;
+        }
+    };
+  
+  template<class TI>
+  struct Policy<TI,TimeSpan,Duration>
+    {
+      static function<TI(TimeSpan const&)>
+      buildChangeHandler (Duration& target)
+        {
+          return bind (Link<TI,Duration>::useLengthAsChange, ref(target), _1 );
+        }
+    };
+  
+  
+  // special treatment for TimeSpan values---------------------------------------
+  
+  template<class TI>
+  struct Policy<TI,Duration,TimeSpan>
+    {
+      static function<TI(Duration const&)>
+      buildChangeHandler (TimeSpan& target)
+        {
+          return bind (Link<TI,TimeSpan>::mutateLength, ref(target), _1 );
+        }
+    };
+  
+  template<>
+  struct Policy<TimeSpan,TimeSpan,TimeSpan>
+    {
+      static function<TimeSpan(TimeSpan const&)>
+      buildChangeHandler (TimeSpan& target)
+        {
+          return bind (Link<TimeSpan,TimeSpan>::mutateTimeSpan, ref(target), _1 );
+        }
+    };
+  
+  
+  
+}}} // namespace lib::time::mutation
+#endif

src/lib/time/control.hpp

@@ -0,0 +1,256 @@
+/*
+  CONTROL.hpp  -  a life time control for feedback and mutation  
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+/** @file control.hpp
+ ** Manipulating and monitoring time entities with life changes.
+ ** This is an control- and callback element to handle any kind of "running"
+ ** time entities. This element is to be provided by the client and then attached
+ ** to the target time entity as a time::Mutation. Internally, a life connection to
+ ** the target is built, allowing both to 
+ ** - to manipulate the target by invoking the function operator
+ ** - to receive change notifications by installing a callback functor.
+ ** 
+ ** The actual type of the changes and modifications is specified as template parameter;
+ ** when later attached to some time entity as a Mutation, the actual changes to be performed
+ ** depend both on this change type and the type of the target time entity (double dispatch).
+ ** The behaviour is similar to applying a static time::Mutation
+ ** 
+ ** \par relevance
+ ** This control element is intended to be used for all kinds of editing and monitoring
+ ** of time-like entities -- be it the running time display in a GUI widget, a ruler marker
+ ** which can be dragged, a modifiable selection or the animated playhead cursor.
+ ** 
+ ** \par usage scenarios
+ ** The time::Control element provides mediating functionality, but doesn't assume or provide
+ ** anything special regarding the usage pattern or the lifecycle, beyond the ability to
+ ** attach listeners, attach to a (different) target and to detach from all connections.
+ ** Especially, no assumptions are made about which side is the server or the client
+ ** and who owns the time::Control element.
+ ** 
+ ** Thus an interface might accept a time::Control element \em reference (e.g. the
+ ** lumiera::Play::Controller uses this pattern) -- meaning that the client owns the
+ ** Control element and might attach listeners, while the implementation (server side)
+ ** will attach the Control to mutate an time value entity otherwise not disclosed
+ ** (e.g. the playhead position of the playback process). Of course, in this case
+ ** the client is responsible for keeping the Control element and all listeners
+ ** alive, and to invoke Control#disconnect prior to destroying the element.
+ ** 
+ ** Of course, the reversed usage situation would be possible as well: an interface
+ ** exposing a time::Control, thus allowing to attach target and listeners, while the
+ ** actual changes will originate somewhere within the service implementation.
+ ** 
+ ** Another usage pattern would be to expose a time::Control \c const&, allowing only to
+ ** impose changes, but not to change the target or listener attachments. To the contrary,
+ ** when exposing only a time::Mutation \c const& through an interface allows only to
+ ** attach new target elements, but not to change listeners or feed any value changes.
+ ** 
+ ** Using time::Control as an implementation building block and just exposing the
+ ** change (function) operators or the listener attachment through an forwarding sub
+ ** interface is another option.
+ ** 
+ ** @note time::Control is default constructible and freely copyable.
+ ** 
+ ** 
+ ** \par changing quantised (grid aligned) time entities
+ ** 
+ ** The time::Control element includes the functionality to handle grid aligned time values,
+ ** both as target and as change/notification value. This ability is compiled in conditionally,
+ ** as including mutation.hpp causes several additional includes, which isn't desirable when
+ ** it comes just to changing plain time values. Thus, to get these additional specialisations,
+ ** the LIB_TIME_TIMEQUQNT_H header guard needs to be defined, which happens automatically
+ ** if lib/time/mutation.hpp is included prior to lib/time/control.hpp.
+ ** 
+ ** \par implementation notes
+ ** - the validity of a given combination of change and target is checked immediately,
+ **   when connecting to the target. Depending on the situation, the actual changes later
+ **   are subject to specific treatment (e.g. frame quantisation)
+ ** - by default time::Control is <b>not threadsafe</b>. But, as each change is basically
+ **   processed within its own call context (function invocation), parallelism is only
+ **   a concern with respect to the value finally visible within the target.
+ ** - the change notification is processed right away, after applying the change to the
+ **   target; of course there is a race between applying the value and building the
+ **   response value passed on as notification. In all cases, the effective change
+ **   notification value is built from the state of the target after applying
+ **   the change, which might or might not reflect the change value passed in.
+ ** 
+ ** @todo include support for QuTimeSpan values                               ////////////////////TICKET #760
+ ** 
+ ** @see TimeControl_test
+ **
+ */
+
+#ifndef LIB_TIME_CONTROL_H
+#define LIB_TIME_CONTROL_H
+
+#include "lib/time/mutation.hpp"
+#include "lib/time/timevalue.hpp"
+#include "lib/time/control-impl.hpp"
+
+
+
+namespace lib {
+namespace time {
+  
+  
+  
+  /**
+   * Frontend/Interface: controller-element to retrieve
+   * and change running time values. time::Control is
+   * a mediator element, which can be attached to some
+   * time value entities as \em mutation, and at the
+   * same time allows to register listeners. When
+   * configured this way, \em changes may be fed
+   * to the function operator(s). These changes
+   * will be imposed to the connected target
+   * and the result propagated to the listeners.
+   * 
+   * @see time::Mutation
+   * @see time::TimeSpan#accept(Mutation const&)
+   */
+  template<class TI>
+  class Control
+    : public mutation::Mutator<TI>
+    {
+      mutation::Propagator<TI> notifyListeners_;
+      
+      virtual void change (Duration&)  const;
+      virtual void change (TimeSpan&)  const;
+      virtual void change (QuTime&)    const;
+        
+    public:
+      void operator() (TI const&)      const;
+      void operator() (Offset const&)  const;
+      void operator() (int)            const;
+      
+      
+      /** install a callback functor to be invoked as notification
+       *  for any changes imposed onto the observed time entity.
+       *  @param toNotify object with \c operator()(TI const&) */
+      template<class SIG>
+      void connectChangeNotification (SIG const& toNotify);
+      
+      /** disconnect from observed entity and
+       *  cease any change notification */
+      void disconnnect();
+    };
+  
+  
+  
+  
+  /* === forward to implementation === */
+  
+  /** impose a new value to the connected target.
+   *  If applicable, the target will afterwards reflect
+   *  that change, and listeners will be notified, passing
+   *  the target's new state.
+   * @throw error::State if not connected to a target
+   * @note the actual change in the target also depends
+   *       on the concrete target type and the type of
+   *       the change. By default, the time value is
+   *       changed; this may include grid alignment.
+   */
+  template<class TI>
+  void
+  Control<TI>::operator () (TI const& newValue)  const
+  {
+    this->ensure_isArmed();
+    notifyListeners_(
+        this->setVal_(newValue));
+  }
+  
+  /** impose an offset to the connected target.
+   *  If applicable, the target will be adjusted by the
+   *  time offset, and listeners will be notified.
+   * @throw error::State if not connected to a target
+   */
+  template<class TI>
+  void
+  Control<TI>::operator () (Offset const& adjustment)  const
+  {
+    this->ensure_isArmed();
+    notifyListeners_(
+        this->offset_(adjustment));
+  }
+  
+  /** nudge the connected target by the given offset steps,
+   *  using either the target's own grid (when quantised),
+   *  or a 'natural' nudge grid
+   * @throw error::State if not connected to a target
+   */
+  template<class TI>
+  void
+  Control<TI>::operator () (int offset_by_steps)  const
+  {
+    this->ensure_isArmed();
+    notifyListeners_(
+        this->nudge_(offset_by_steps));
+  }
+  
+  
+  template<class TI>
+  void
+  Control<TI>::disconnnect()
+  {
+    notifyListeners_.disconnect();
+    this->unbind();
+  }
+  
+  template<class TI>
+  template<class SIG>
+  void
+  Control<TI>::connectChangeNotification (SIG const& toNotify)
+  {
+    if (this->offset_)
+      { // we're already connected: thus propagate current value
+        TI currentValue = this->offset_(Offset::ZERO);
+        toNotify (currentValue);
+      }
+    notifyListeners_.attach (toNotify);
+  }
+  
+  
+  /* ==== Implementation of the Mutation interface ==== */
+  
+  template<class TI>
+  void
+  Control<TI>::change (Duration& targetDuration)  const
+  {
+    this->bind_to (targetDuration);
+  }
+  
+  template<class TI>
+  void
+  Control<TI>::change (TimeSpan& targetInterval)  const
+  {
+    this->bind_to (targetInterval);
+  }
+  
+  template<class TI>
+  void
+  Control<TI>::change (QuTime& targetQuTime)  const
+  {
+    this->bind_to (targetQuTime);
+  }
+  
+}} // lib::time
+#endif

src/lib/time/lumitime.cpp

@@ -41,6 +41,9 @@ namespace time {
   const Time Time::MIN ( TimeValue::buildRaw_(-_raw(Time::MAX)                             ) );
   const Time Time::ZERO;
   
+  const Offset Offset::ZERO (Time::ZERO);
+  
+  
   
   /** convenience constructor to build an
    *  internal Lumiera Time value from the usual parts
@@ -116,14 +119,21 @@ namespace time {
   }
   
   
-  /** duration of the given number of frames */
-  Duration::Duration (ulong count, FrameRate const& fps)
-    : Offset(TimeValue (count? lumiera_frame_duration (fps/count) : _raw(Duration::NIL)))
+  /** offset by the given number of frames. */
+  Offset::Offset (int64_t count, FrameRate const& fps)
+    : TimeValue (count?  (count<0? -1:+1) * lumiera_framecount_to_time (::abs(count), fps)
+                      : _raw(Duration::NIL))
+    { }
+  
+  /** duration of the given number of frames.
+   * @note always positive; count used absolute */
+  Duration::Duration (int64_t count, FrameRate const& fps)
+    : TimeValue (count? lumiera_framecount_to_time (abs(count), fps) : _raw(Duration::NIL))
     { }
   
   
   /** constant to indicate "no duration" */
-  const Duration Duration::NIL = Offset(TimeValue(0));
+  const Duration Duration::NIL (Time::ZERO);
   
   
   

src/lib/time/mutation.cpp

@@ -48,17 +48,6 @@ namespace time {
   Mutation::~Mutation()   { }  // emit VTable here....
   
   
-  /** @internal actually force a change
-   *  into a target time entity to mutate.
-   *  Mutation is declared fried to TimeValue
-   *  and thus is allowed to influence the basic
-   *  value stored in each time entity
-   */
-  void
-  Mutation::imposeChange (TimeValue& target, TimeValue const& valueToSet)
-  {
-    target = valueToSet; 
-  }
   
   
   
@@ -157,14 +146,14 @@ namespace time {
       virtual void
       change (Duration& target) const
         {
-          imposeChange (target, TimeVar(target)+=adjustment_);
+          imposeChange (target, adjustment_);
         }
       
       
       virtual void
       change (TimeSpan& target) const
         {
-          imposeChange (target, TimeVar(target)+=adjustment_);
+          imposeChange (target, adjustment_);
         }
       
       
@@ -173,7 +162,7 @@ namespace time {
       virtual void
       change (QuTime& target)   const
         {
-          imposeChange (target, TimeVar(target)+=adjustment_);
+          imposeChange (target, adjustment_);
         }
       
       
@@ -205,7 +194,7 @@ namespace time {
   /** 
    * concrete time value mutation:
    * nudge target value by the given number of 'steps',
-   * relative to either the given grid.
+   * relative to the given grid.
    */
   class NudgeMutation
     : public ImposeOffsetMutation
@@ -234,6 +223,8 @@ namespace time {
    * @note currently the natural grid is hard wired,
    *       just interpreting the step parameter as
    *       offset in seconds.
+   * @see mutation#imposeChange (TimeValue, int)
+   * @see mutation#imposeChange (QuTime, int)
    */
   class NaturalNudgeMutation
     : public ClonableMutation
@@ -243,31 +234,22 @@ namespace time {
       virtual void
       change (Duration& target) const
         {
-          imposeChange (target, TimeVar(target)+=Time(FSecs(steps_)));
+          imposeChange (target, steps_);
         }
       
       
       virtual void
       change (TimeSpan& target) const
         {
-          imposeChange (target, TimeVar(target)+=Time(FSecs(steps_)));
+          imposeChange (target, steps_);
         }
       
       
-      /** Special treatment: use the quantised time's own grid;
-       *  retrieve the corresponding grid point, offset it by the step-parameter,
-       *  then retrieve the corresponding time from the quantised time's
-       *  underlying quantiser (grid).
-       * @note when the #steps_ parameter is zero, what happens here effectively
-       *       is the materialisation of the quantised target time, i.e. making
-       *       the quantisation explicit and storing the resulting value. */
+      /** @note special treatment: use the quantised time's own grid */
       virtual void
       change (QuTime& target)   const
         {
-          PQuant const& grid (target);
-          int64_t originalGridPoint = grid->gridPoint(target);
-          int64_t adjustedGridPoint = originalGridPoint + steps_;
-          imposeChange (target, grid->timeOf (adjustedGridPoint));
+          imposeChange (target, steps_);
         }
       
    

src/lib/time/mutation.hpp

@@ -112,7 +112,10 @@ namespace time {
       static EncapsulatedMutation nudge (int adjustment, Symbol gridID);
       
     protected:
-      static void imposeChange (TimeValue&, TimeValue const&);
+      static TimeValue& imposeChange (TimeValue&, TimeValue const&);
+      static TimeValue& imposeChange (TimeValue&, Offset const&);
+      static TimeValue& imposeChange (TimeValue&, int);
+      static TimeValue& imposeChange (QuTime&, int);
     };
   
   
@@ -126,5 +129,53 @@ namespace time {
 #endif
   
   
+  
+  /* === implementing the actual changes === */
+  
+  /** @internal actually force a change into a target time entity to mutate.
+   *  Mutation is declared fried to TimeValue and thus is allowed to influence
+   *  the basic value stored in each time entity
+   */
+  inline TimeValue&
+  Mutation::imposeChange (TimeValue& target, TimeValue const& valueToSet)
+  {
+    return target = valueToSet; 
+  }
+  
+  /** @internal variation to mutate a target time value by applying an offset */
+  inline TimeValue&
+  Mutation::imposeChange (TimeValue& target, Offset const& offset)
+  {
+    return imposeChange (target, TimeVar(target) += offset); 
+  }
+  
+  /** @internal nudge a target time value by a step wise offset.
+   *  The standard case uses a fixed offset of 1 second per step                  //////////////////TICKET #810
+   */
+  inline TimeValue&
+  Mutation::imposeChange (TimeValue& target, int steps)
+  {
+    return imposeChange (target, TimeVar(target) += Time(FSecs(steps)));
+  }
+  
+#ifdef LIB_TIME_TIMEQUQNT_H
+  /** @internal Special treatment for quantised target values:
+   *  use the quantised time's own grid; retrieve the corresponding grid point,
+   *  offset it by the step-parameter, then retrieve the corresponding time from
+   *  the quantised time's underlying quantiser (grid) and impose that as change.
+   * @note when the #steps parameter is zero, what happens here effectively
+   *       is the materialisation of the quantised target time, i.e. making
+   *       the quantisation explicit and storing the resulting value. */
+  inline TimeValue&
+  Mutation::imposeChange (QuTime& target, int steps)
+  {
+    PQuant const& grid (target);
+    int64_t originalGridPoint = grid->gridPoint(target);
+    int64_t adjustedGridPoint = originalGridPoint + steps;
+    return imposeChange (target, grid->timeOf (adjustedGridPoint));
+  }
+#endif
+  
+  
 }} // lib::time
 #endif

src/lib/time/quantiser.cpp

@@ -51,10 +51,19 @@ namespace time {
                            , LUMIERA_ERROR_UNKNOWN_GRID);
       return grid_found;
     }
-    
   }//(End) implementation helpers
   
   
+  PQuant
+  getDefaultGridFallback()
+  {
+    static PQuant globalDefaultGrid (new FixedFrameQuantiser(1));
+    return globalDefaultGrid;                                                    ///////////////////////TICKET #810
+  };
+  
+  
+  
+  
   
   Grid::~Grid() { } // hint to emit the VTable here...
   
@@ -191,7 +200,7 @@ namespace time {
   {
     Time gt(gridTime);
     TimeVar timePoint = gt + origin_;
-    timePoint += gridOffset * raster_;
+    timePoint += gridOffset * Offset(raster_);
     return timePoint;
   }
   

src/lib/time/quantiser.hpp

@@ -64,6 +64,17 @@ namespace time {
   }
   
   
+  
+  /** @todo placeholder for accessing
+   *        a current or default session grid.
+   *        To be implemented later.
+   */
+  PQuant getDefaultGridFallback();                                        ///////////////////////TICKET #810
+  
+  
+  
+  
+  
   /**
    * Facility to create grid-aligned time values.
    * Effectively, a quantiser exposes the value Grid API, but

src/lib/time/timevalue.hpp

@@ -42,6 +42,11 @@ namespace time {
   
   namespace error = lumiera::error;
   
+  // forwards...
+  class FrameRate;
+  class TimeSpan;
+  class Mutation;
+  
   
   /**
    * basic constant internal time value.
@@ -193,6 +198,11 @@ namespace time {
         : TimeValue(TimeVar(target) -= origin)
         { }
       
+      Offset (int64_t count, FrameRate const& fps);
+      
+      static const Offset ZERO;
+      
+      
       TimeValue
       abs()  const
         {
@@ -241,7 +251,6 @@ namespace time {
    * @warning do not mix up gavl_time_t and FSecs */
   typedef boost::rational<long> FSecs;
   
-  class FrameRate;
   
   /**
    * Lumiera's internal time value datatype.
@@ -304,8 +313,6 @@ namespace time {
   
   
   
-  class TimeSpan;
-  class Mutation;
   
   /**
    * Duration is the internal Lumiera time metric.
@@ -315,33 +322,63 @@ namespace time {
    * possibility to send a \em Mutation message.
    */
   class Duration
-    : public Offset
+    : public TimeValue
     {
       /// direct assignment prohibited
       Duration& operator= (Duration const&);
       
     public:
       Duration (Offset const& distance)
-        : Offset(distance.abs())
+        : TimeValue(distance.abs())
         { }
       
       explicit
       Duration (TimeValue const& timeSpec)
-        : Offset(Offset(timeSpec).abs())
+        : TimeValue(Offset(timeSpec).abs())
         { }
       
       explicit
       Duration (FSecs const& timeSpan_in_secs)
-        : Offset(Offset(Time(timeSpan_in_secs)).abs())
+        : TimeValue(Offset(Time(timeSpan_in_secs)).abs())
         { }
       
       Duration (TimeSpan const& interval);
-      Duration (ulong count, FrameRate const& fps);
+      Duration (int64_t count, FrameRate const& fps);
       
       static const Duration NIL;
       
       void accept (Mutation const&);
+      
+      /// Supporting backwards use as offset
+      Offset operator- ()  const;
+      
     };
+    
+  //-- support using a Duration to build offsets ---------------
+  
+  inline Duration
+  operator+ (Duration const& base, Duration const& toAdd)
+  {
+    return Offset(base) + Offset(toAdd);
+  }
+  
+  inline Offset
+  operator* (int factor, Duration const& dur)
+  {
+    return factor * Offset(dur);
+  }
+  
+  inline Offset
+  operator* (Duration const& dur, int factor)
+  {
+    return factor*dur;
+  }
+  
+  inline Offset
+  Duration::operator- ()  const
+  {
+    return -1 * (*this); 
+  }
   
   
   
@@ -493,7 +530,7 @@ namespace time {
   
   inline
   Duration::Duration (TimeSpan const& interval)
-    : Offset(interval.duration())
+    : TimeValue(interval.duration())
     { }
   
   inline

src/lib/typed-allocation-manager.hpp

@@ -63,8 +63,8 @@
 
 
 
-#ifndef CONTROL_TYPED_ALLOCATION_MANAGER_H
-#define CONTROL_TYPED_ALLOCATION_MANAGER_H
+#ifndef LIB_TYPED_ALLOCATION_MANAGER_H
+#define LIB_TYPED_ALLOCATION_MANAGER_H
 
 //#include "pre.hpp"
 #include "lib/error.hpp"

src/lumiera/main.cpp

@@ -37,15 +37,15 @@
 using lib::Cmdline;
 using lumiera::Subsys;
 using lumiera::AppState;
-using lumiera::ON_GLOBAL_INIT;
 
 namespace {
   Subsys& engine  = backend::EngineFacade::getDescriptor();
   Subsys& netNode = backend::NetNodeFacade::getDescriptor();
   Subsys& script  = backend::ScriptRunnerFacade::getDescriptor();
-  Subsys& player  = lumiera::DummyPlayer::getDescriptor();
+  Subsys& player  = lumiera::DummyPlayer::getDescriptor();        ///////TODO: just a dummy, until we're able to render
   Subsys& builder = proc::Facade::getBuilderDescriptor();
   Subsys& session = proc::Facade::getSessionDescriptor();
+  Subsys& playOut = proc::Facade::getPlayOutDescriptor();
   Subsys& lumigui = gui::GuiFacade::getDescriptor();
 }
 
@@ -63,16 +63,20 @@ main (int argc, const char* argv[])
       lumiera::Option options (args);
       application.init (options);
       
-      session.depends (builder);
+//    session.depends (builder);
       netNode.depends (session);
       netNode.depends (engine);
+//    playOut.depends (engine);
+//    playOut.depends (session);
 //    lumigui.depends (session);   //////TODO commented out in order to be able to start up a dummy GuiStarterPlugin
 //    lumigui.depends (engine);
+      player.depends (playOut);    //////TODO dummy player, until we're able to render
       lumigui.depends (player);
       script.depends (session);
       script.depends (engine);
       
       application.maybeStart (session);
+      application.maybeStart (playOut);
       application.maybeStart (netNode);
       application.maybeStart (lumigui);
       application.maybeStart (script);

src/proc/asset/entry-id.hpp

@@ -48,6 +48,7 @@
 #include "lib/hash-indexed.hpp"
 #include "lib/util.hpp"
 
+#include <boost/functional/hash.hpp>
 #include <boost/operators.hpp>
 #include <iostream>
 #include <string>

src/proc/asset/viewer.cpp

@@ -0,0 +1,70 @@
+/*
+  Viewer  -  asset corresponding to a viewer element in the GUI
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+* *****************************************************/
+
+
+#include "proc/asset/viewer.hpp"
+//#include "proc/mobject/session/track.hpp"
+//#include "proc/mobject/placement.hpp"
+//#include "proc/mobject/session/mobjectfactory.hpp"
+//#include "proc/mobject/session/binding.hpp"
+//#include "proc/assetmanager.hpp"
+
+
+namespace asset {
+  
+//  using lib::AutoRegistered;
+  
+  
+  
+  /** @todo anything significant to do here??? */
+  Viewer::Viewer (const Asset::Ident& idi)
+    : Struct (idi)
+  {
+    UNIMPLEMENTED ("anything regarding Viewer Assets");
+  }
+  
+  
+//PViewer
+//Viewer::create (Asset::Ident const& idi)
+//{
+//  REQUIRE (getRegistry, "can't create a Timeline prior to session initialisation");
+//  
+//  PTimeline newElement (AssetManager::instance().wrap (*new Viewer(idi)));
+//  getRegistry().append (newElement);
+//  
+//  ENSURE (newElement);
+//  ENSURE (getRegistry().isRegistered (*newElement));
+//  return newElement;
+//}
+  
+  
+  void
+  Viewer::unlink ()
+  {
+//  AutoRegistered<Timeline>::detach();
+    Struct::unlink();
+  }
+  
+  
+  
+  
+} // namespace asset

src/proc/asset/viewer.hpp

@@ -0,0 +1,107 @@
+/*
+  VIEWER.hpp  -  asset corresponding to a viewer element in the GUI
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+/** @file viewer.hpp
+ ** structural element corresponding to a viewer in the GUI.
+ ** This Asset marks an attachment point, allowing other (output producing) 
+ ** elements to be connected to ("viewer attachment"). The typical standard case
+ ** is a Timeline: It features a set of global pipes (busses), which collect all
+ ** produced data. Yet without an explicit output connection, this data can't be
+ ** generated, because in Lumiera, actual output is always retrieved ("pulled")
+ ** starting from an output sink. Thus, in order to \em perform (play, render)
+ ** the timeline, an "view connection" needs to be established: this connection
+ ** is represented by an session::BindingMO, linking the Timeline to an
+ ** asset::Viewer. Consequently, the same output mapping and translation
+ ** mechanism used for Sequence-Timeline- (and VirtualClip-)Bindings
+ ** is employed in this situation to figure out the real output port.
+ ** 
+ ** @todo WIP-WIP-WIP as of 5/11
+ ** 
+ ** @see Session
+ ** @see Timeline
+ ** @see StructFactory
+ **
+ */
+
+
+#ifndef ASSET_VIEWER_H
+#define ASSET_VIEWER_H
+
+#include "proc/asset/struct.hpp"
+//#include "proc/mobject/mobject.hpp"
+//#include "proc/mobject/placement.hpp"
+#include "proc/mobject/mobject-ref.hpp"
+//#include "proc/mobject/session/binding.hpp"         ////TODO avoidable??
+#include "lib/p.hpp"
+#include "lib/element-tracker.hpp"
+
+
+//#include <vector>
+//#include <string>
+
+//using std::vector;
+//using std::string;
+
+namespace mobject {
+namespace session {
+  
+  class Binding;
+  typedef MORef<Binding> RBinding;
+}}
+
+
+namespace asset {
+  
+  
+  using lumiera::P;
+  class Viewer;
+  typedef lumiera::P<Viewer> PViewer;
+  
+  
+  /**
+   * TODO type comment
+   */
+  class Viewer
+    : public Struct
+//  , public lib::AutoRegistered<Viewer>
+    {
+      
+      Viewer (Ident const&); /////TODO ctor params????
+      
+    public:
+      /** create and register a new Timeline instance */
+//    static PTimeline create (Asset::Ident const& idi, RBinding const& sequenceBinding);
+      
+    protected:
+      virtual void unlink ();
+      
+    };
+    
+  
+  
+  
+///////////////////////////TODO currently just fleshing the API
+  
+  
+} // namespace asset
+#endif

src/proc/control/command-argument-holder.hpp

@@ -54,7 +54,6 @@
 
 namespace control {
   
-  using lib::TypedAllocationManager;
   using lib::InPlaceBuffer;
   using std::string;
   

src/proc/control/command-def.hpp

@@ -67,7 +67,7 @@
 #include "lib/bool-checkable.hpp"
 #include "lib/meta/function.hpp"
 #include "lib/meta/typelist.hpp"
-#include "lib/meta/typelist-util.hpp"
+#include "lib/meta/typelist-manip.hpp"
 #include "lib/meta/tuple.hpp"
 #include "lib/util.hpp"
 

src/proc/control/command-impl.hpp

@@ -57,7 +57,6 @@
 
 namespace control {
   
-  using lib::TypedAllocationManager;
   using std::tr1::function;
   using std::tr1::shared_ptr;
   

src/proc/control/command-signature.hpp

@@ -30,7 +30,7 @@
  ** we need to accept functor objects with a very specific and predetermined
  ** signature, thus allowing for strict type checking by the compiler.
  ** 
- ** \par Relation of function signatures
+ ** \par Relation of function signatures (MEM = type of the "memento" for Undo)
  ** - operation: void(P1,..PN)
  ** - captureUndo: MEM(P1,..PN)
  ** - undoOperation void(P1,..PN,MEM)
@@ -47,14 +47,11 @@
 #define CONTROL_COMMAND_SIGNATURE_H
 
 //#include "pre.hpp"
-//#include "lib/symbol.hpp"
 #include "lib/meta/function.hpp"
 #include "lib/meta/typelist.hpp"
-#include "lib/meta/typelist-util.hpp"
+#include "lib/meta/typelist-manip.hpp"
 #include "lib/meta/typeseq-util.hpp"
-//#include "lib/meta/tuple.hpp"
 
-//#include <tr1/memory>
 #include <tr1/functional>
 
 
@@ -62,24 +59,20 @@
 
 namespace control {
   
-//  using lib::Symbol;
-//  using std::tr1::shared_ptr;
   using std::tr1::function;
   
   using lumiera::typelist::FunctionSignature;
   using lumiera::typelist::FunctionTypedef;
   using lumiera::typelist::Types;
-//using lumiera::typelist::NullType;
-//using lumiera::typelist::Tuple;
   using lumiera::typelist::Append;
   using lumiera::typelist::SplitLast;
   
-
+  
   /** 
-   * Metaprogramming helper for building Command function signatures. 
+   * Metaprogramming helper for building Command function signatures.
    * The complete definition context of any command is templated to the signature
    * of the actual command operation and to the memento type. The typedefs embedded
-   * within CommandSignature<SIG,MEM> allows accepting suitable typed functions
+   * within CommandSignature<SIG,MEM> allows for accepting suitable typed functions
    * to implement the command in question.
    */
   template<typename SIG, typename MEM>
@@ -93,7 +86,7 @@ namespace control {
       
     public:
       typedef typename FunctionTypedef<void, Args>::Sig          OperateSig;
-      typedef typename FunctionTypedef<MEM, Args>::Sig           CaptureSig;
+      typedef typename FunctionTypedef<MEM,  Args>::Sig          CaptureSig;
       typedef typename FunctionTypedef<void, ExtendedArgs>::Sig  UndoOp_Sig;
       typedef Args                                               CmdArgs;
       typedef MEM                                                Memento;
@@ -103,16 +96,23 @@ namespace control {
   
   
   /** 
-   * Type analysis helper template. 
+   * Type analysis helper template.
    * Used for dissecting a given type signature to derive
    * the related basic operation signature, the signature of a possible Undo-function
    * and the signature necessary for capturing undo information. The implementation
    * relies on re-binding an embedded type defining template, based on the actual
    * case, as identified by the structure of the given parameter signature.
+   * 
+   * To use this template, it is instantiated with the signature of a functor object
+   * in question. Depending on the actual situation, the compiler will then either
+   * pick Case1 or Case2 -- thus allowing the client in any case to pick up the
+   * correct signatures for Operation, Capture and Undo-function from the
+   * public typedefs within \c UndoSignature
    */
   template<typename SIG>
   class UndoSignature
     {
+      // preparation:  dissect the function signature into arguments and result
       typedef typename FunctionSignature< function<SIG> >::Args Args;
       typedef typename FunctionSignature< function<SIG> >::Ret  Ret;
       

src/proc/control/handling-pattern.hpp

@@ -67,6 +67,7 @@ namespace control {
    * It is returned when invoking a HandlingPattern
    * and can be used to check for success and/or re-throw
    * any Exception encountered during the command execution.
+   * @todo couldn't that be replaced by a lib::Result<void> instance??
    */
   class ExecResult
     : public lib::BoolCheckable<ExecResult>

src/proc/engine/buffer-metadata.hpp

@@ -0,0 +1,584 @@
+/*
+  BUFFER-METADATA.hpp  -  internal metadata for data buffer providers
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+/** @file buffer-metadate.hpp
+ ** Metadata for managing and accessing buffers. The Lumiera Engine uses the
+ ** Abstraction of an BufferProvider to handle various kinds of buffer organisation
+ ** and access in a uniform way. Actually, buffers can be exposed and provided by several
+ ** facilities, which might even be implemented through an external library. Thus the engine
+ ** and the abstraction placed in between needs a common set of control data, to be able to
+ ** expose the correct buffer for each request. Typically -- and independent of the actual
+ ** implementation -- the following properties need to be tracked
+ ** - that overall storage size available within the buffer
+ ** - a pair of custom \em creator and \em destructor functions to use together with this buffer
+ ** - an additional client key to distinguish otherwise otherwise identical client requests
+ ** These three distinctions are applied in sequence, thus forming a tree with 3 levels.
+ ** Only the first distinguishing level (the size) is mandatory. The others are provided,
+ ** because some of the foreseeable buffer providers allow to re-access the data placed
+ ** into the buffer, by assigning an internally managed ID to the buffer. The most
+ ** prominent example is the frame cache, which obviously needs to keep track of
+ ** the buffers after the render engine is finished, while the engine code
+ ** just accesses yet another buffer to place the results of calculations.
+ ** 
+ ** These additional distinctions and properties are associated with the help of the
+ ** BufferDescriptor, embedded into each BuffHandle. While the engine just uses these
+ ** handles in the way of a pointer, the buffer descriptor acts as an additional tag
+ ** attached to the buffer access, allowing to re-access a context within the
+ ** buffer provider implementation.
+ ** 
+ ** @see buffer-provider.hpp
+ ** @see BufferMetadata_test
+ ** @see BufferProviderProtocol_test
+ */
+
+#ifndef PROC_ENGINE_BUFFR_METADATA_H
+#define PROC_ENGINE_BUFFR_METADATA_H
+
+
+#include "lib/error.hpp"
+#include "lib/symbol.hpp"
+#include "lib/functor-util.hpp"
+
+#include <tr1/functional>
+#include <boost/functional/hash.hpp>
+#include <boost/noncopyable.hpp>
+
+
+namespace engine {
+  
+  using lib::HashVal;
+  using lib::Literal;
+  using std::tr1::bind;
+  using std::tr1::function;
+  using std::tr1::placeholders::_1;
+  
+  namespace error = lumiera::error;
+  
+  namespace metadata {
+    class Key;
+    class Entry;
+  }
+  
+  
+  
+  enum BufferState
+    { NIL,
+      FREE,
+      LOCKED,
+      EMITTED,
+      BLOCKED
+    };
+  
+  
+  /**
+   * an opaque ID to be used by the BufferProvider implementation.
+   * Typically this will be used, to set apart some pre-registered
+   * kinds of buffers. It is treated as being part of the buffer type.
+   * LocalKey objects may be copied but not re-assigned or changed.
+   */
+  class LocalKey
+    {
+      uint64_t privateID_;
+      
+    public:
+      LocalKey (uint64_t opaqueValue=0)
+        : privateID_(opaqueValue)
+        { }
+      
+      operator uint64_t()  const { return privateID_; }
+      
+      friend size_t
+      hash_value (LocalKey const& lkey)
+      {
+        boost::hash<uint64_t> hashFunction;
+        return hashFunction(lkey.privateID_);
+      }
+      
+    private:
+      /** assignment usually prohibited */
+      LocalKey& operator= (LocalKey const& o)
+        {
+          privateID_ = o.privateID_;
+          return *this;
+        }
+      
+      /** but Key assignments are acceptable */
+      friend class metadata::Key;
+    };
+  
+  
+  namespace { // Helpers for construction within the buffer...
+    
+    template<class X>
+    inline void
+    buildIntoBuffer (void* storageBuffer)
+    {
+      new(storageBuffer) X();
+    }
+    
+    template<class X, typename A1>
+    inline void
+    buildIntoBuffer_A1 (void* storageBuffer, A1 arg1)
+    {
+      new(storageBuffer) X(arg1);
+    }
+    
+    template<class X>
+    inline void
+    destroyInBuffer (void* storageBuffer)
+    {
+      X* embedded = static_cast<X*> (storageBuffer);
+      embedded->~X();
+    }
+  }//(End)placement-new helpers
+  
+  
+  /**
+   * A pair of functors to maintain a datastructure within the buffer.
+   * TypeHandler describes how to outfit the buffer in a specific way.
+   * When defined, the buffer will be prepared when locking and cleanup
+   * will be invoked automatically when releasing. Especially, this
+   * can be used to \em attach an object to the buffer (placement-new) 
+   */
+  struct TypeHandler
+    {
+      typedef function<void(void*)> DoInBuffer;
+      
+      DoInBuffer createAttached;
+      DoInBuffer destroyAttached;
+      
+      /** build an invalid NIL TypeHandler */
+      TypeHandler()
+        : createAttached()
+        , destroyAttached()
+        { }
+      
+      /** build a TypeHandler
+       *  binding to arbitrary constructor and destructor functions.
+       *  On invocation, these functions get a void* to the buffer.
+       * @note the functor objects created from these operations
+       *       might be shared for handling multiple buffers.
+       *       Be careful with any state or arguments.
+       */
+      template<typename CTOR, typename DTOR>
+      TypeHandler(CTOR ctor, DTOR dtor)
+        : createAttached (ctor)
+        , destroyAttached (dtor)
+        { }
+      
+      /** builder function defining a TypeHandler
+       *  to place a default-constructed object
+       *  into the buffer. */
+      template<class X>
+      static TypeHandler
+      create ()
+        {
+          return TypeHandler (buildIntoBuffer<X>, destroyInBuffer<X>);
+        }
+      
+      template<class X, typename A1>
+      static TypeHandler
+      create (A1 a1)
+        {
+          return TypeHandler ( bind (buildIntoBuffer_A1<X,A1>, _1, a1)
+                             , destroyInBuffer<X>);
+        }
+      
+      bool
+      isValid()  const
+        {
+          return bool(createAttached)
+              && bool(destroyAttached);
+        }
+      
+      friend HashVal
+      hash_value (TypeHandler const& handler)
+      {
+        HashVal hash(0);
+        if (handler.isValid())
+          {
+            boost::hash_combine(hash, handler.createAttached);
+            boost::hash_combine(hash, handler.destroyAttached);
+          }
+        return hash;
+      }
+      
+      friend bool
+      operator== (TypeHandler const& left, TypeHandler const& right)
+      {
+        return (!left.isValid() && !right.isValid())
+            || (  util::rawComparison(left.createAttached, right.createAttached)
+               && util::rawComparison(left.destroyAttached, right.destroyAttached)
+               );
+      }
+      friend bool
+      operator!= (TypeHandler const& left, TypeHandler const& right)
+      {
+        return !(left == right);       
+      }
+    };
+  
+  
+  namespace { // internal constants to mark the default case
+    
+    const LocalKey UNSPECIFIC;
+    const TypeHandler RAW_BUFFER;
+    
+    inline bool
+    nontrivial (TypeHandler const& toVerify)
+    {
+      return RAW_BUFFER != toVerify;
+    }
+    
+    inline bool
+    nontrivial (LocalKey const& toVerify)
+    {
+      return UNSPECIFIC != toVerify;
+    }
+  }
+  
+  
+  
+  /* === Implementation === */
+  
+  namespace metadata {
+    
+    using error::LUMIERA_ERROR_LIFECYCLE;
+    using error::LUMIERA_ERROR_BOTTOM_VALUE;
+    
+    namespace { // details of hash calculation
+        template<typename VAL>
+        HashVal
+        chainedHash(HashVal accumulatedHash, VAL changedValue)
+        {
+          boost::hash_combine (accumulatedHash, changedValue);
+          return accumulatedHash;
+        }
+    }
+    
+    class Key
+      {
+        HashVal parent_;
+        HashVal hashID_;
+        
+      protected:
+        size_t storageSize_;
+        TypeHandler instanceFunc_;
+        LocalKey specifics_;
+        
+        
+      public:
+        /** build a standard basic key describing a kind of Buffer. 
+         * @param familyID basic hash seed value to distinguish
+         *                 families of buffer types managed by
+         *                 different BufferProvider instances
+         * @param storageSize fundamental info: buffer size
+         */
+        Key (HashVal familyID, size_t storageSize)
+          : parent_(familyID)
+          , hashID_(chainedHash (familyID, storageSize))
+          , storageSize_(storageSize)
+          , instanceFunc_(RAW_BUFFER)
+          , specifics_(UNSPECIFIC)
+          { }
+        
+        // standard copy operations permitted
+        
+        /** create a derived buffer type description.
+         *  Using a different storage size than the parent type,
+         *  all else remaining the same
+         */
+        Key (Key const& parent, size_t differingStorageSize) 
+          : parent_(parent.hashID_)
+          , hashID_(chainedHash (parent_, differingStorageSize))
+          , storageSize_(differingStorageSize)  // differing from parent
+          , instanceFunc_(parent.instanceFunc_)
+          , specifics_(parent.specifics_)
+          { }
+        
+        
+        /** create a derived buffer type description.
+         *  Using different ctor and dtor functions,
+         *  all else remaining the same as with parent
+         */
+        Key (Key const& parent, TypeHandler const& differingTypeHandlerFunctions) 
+          : parent_(parent.hashID_)
+          , hashID_(chainedHash (parent_, differingTypeHandlerFunctions))
+          , storageSize_(parent.storageSize_)
+          , instanceFunc_(differingTypeHandlerFunctions)  // differing from parent
+          , specifics_(parent.specifics_)
+          { }
+        
+        
+        /** create a derived buffer type description.
+         *  Using a different private ID than the parent type,
+         *  all else remaining the same
+         */
+        Key (Key const& parent, LocalKey anotherTypeSpecificInternalID) 
+          : parent_(parent.hashID_)
+          , hashID_(chainedHash (parent_, anotherTypeSpecificInternalID))
+          , storageSize_(parent.storageSize_)
+          , instanceFunc_(parent.instanceFunc_)
+          , specifics_(anotherTypeSpecificInternalID)  // differing from parent
+          { }
+        
+        
+        HashVal parentKey()  const { return parent_;}
+        operator HashVal()   const { return hashID_;}
+      };
+    
+    
+    class Entry
+      : public Key
+      {
+        BufferState state_;
+        const void* buffer_;
+        
+      public:
+        virtual BufferState
+        state()  const
+          {
+            __must_not_be_NIL();
+            return state_;
+          }
+        
+        virtual const void*
+        access()  const
+          {
+            __must_not_be_NIL();
+            __must_not_be_FREE();
+            
+            ENSURE (buffer_);
+            return buffer_;
+          }
+        
+        virtual Entry&
+        mark (BufferState newState)
+          {
+            switch (this->state_)
+              {
+              case NIL:  __must_not_be_NIL();
+              case FREE: __must_not_be_FREE();
+                
+              case LOCKED:
+                if (newState == EMITTED) break; // allow transition
+                
+              case EMITTED:
+                if (newState == BLOCKED) break; // allow transition
+                
+              case BLOCKED:
+                if (newState == FREE)           // note fall through for LOCKED and EMITTED too
+                  {
+                    buffer_ = 0;
+                    break; // allow transition
+                  }
+              default:
+                throw error::Fatal ("Invalid buffer state encountered.");
+              }
+            state_ = newState;
+            return *this;
+          }
+        
+        
+        void
+        __must_not_be_NIL()  const
+          {
+            if (NIL == state_)
+              throw error::Fatal ("Concrete buffer entry with state==NIL encountered."
+                                  "State transition logic broken (programming error)");
+          }
+        
+        void
+        __must_not_be_FREE()  const
+          {
+            if (FREE == state_)
+                throw error::Logic ("Buffer is inaccessible (marked as free). "
+                                    "Need a new buffer pointer in order to lock an entry. "
+                                    "You should invoke markLocked(buffer) prior to access."
+                                   , LUMIERA_ERROR_LIFECYCLE );
+          }
+      };
+  }
+  
+  
+  
+  
+  
+  class Metadata
+    : boost::noncopyable
+    {
+      Literal id_;
+      HashVal family_;
+      
+    public:
+      
+      typedef metadata::Key Key;
+      typedef metadata::Entry Entry;
+      
+      /** establish a metadata registry.
+       *  Such will maintain a family of buffer type entries
+       *  and provide a service for storing and retrieving metadata
+       *  for concrete buffer entries associated with these types.
+       * @param implementationID to distinguish families
+       *        of type keys belonging to different registries.
+       */
+      Metadata (Literal implementationID)
+        : id_(implementationID)
+        , family_(hash_value(id_))
+        { }
+      
+      /** combine the distinguishing properties
+       *  into a single type key, which will be known/remembered
+       *  from that point on. Properties are combined according to
+       *  a fixed type specialisation order, with the buffer size
+       *  forming the base level, possible TypeHandler functors the
+       *  second level, and implementation defined LocalKey entries
+       *  the third level. All these levels describe abstract type
+       *  keys, not entries for concrete buffers. The latter are
+       *  always created as children of a known type key.  
+       */
+      Key
+      key ( size_t storageSize
+          , TypeHandler instanceFunc =RAW_BUFFER
+          , LocalKey specifics =UNSPECIFIC)
+        {
+          REQUIRE (storageSize);
+          Key typeKey = trackKey (family_, storageSize);
+          
+          if (nontrivial(instanceFunc))
+            {
+              typeKey = trackKey (typeKey, instanceFunc);
+            }
+          
+          if (nontrivial(specifics))
+            {
+              typeKey = trackKey (typeKey, specifics);
+            }
+          
+          return typeKey;
+        }
+      
+      /** create a sub-type, using a different type/handler functor */
+      Key
+      key (Key const& parentKey, TypeHandler const& instanceFunc)
+        {
+          return trackKey (parentKey, instanceFunc);
+        }
+      
+      /** create a sub-type, using a different private-ID (implementation defined) */
+      Key
+      key (Key const& parentKey, LocalKey specifics)
+        {
+          return trackKey (parentKey, specifics);
+        }
+      
+      Key
+      key (Key const& parentKey, const void* concreteBuffer)
+        {
+          UNIMPLEMENTED ("create sub-object key for concrete buffer");
+        }
+      
+      Key const&
+      get (HashVal hashID)
+        {
+          UNIMPLEMENTED ("access the plain key entry");
+        }
+      
+      Entry&
+      get (Key key)
+        {
+          UNIMPLEMENTED ("access, possibly create metadata records");
+        }
+      
+      bool
+      isKnown (HashVal key)  const
+        {
+          UNIMPLEMENTED ("diagnostics: known record?");
+        }
+      
+      bool
+      isLocked (HashVal key)  const
+        {
+          UNIMPLEMENTED ("diagnostics: actually locked buffer instance record?");
+        }
+      
+      
+      /* == memory management == */
+      
+      Entry& markLocked (Key const& parentKey, const void* buffer);
+      void release (HashVal key);
+      
+    private:
+            
+      template<typename PAR, typename DEF>
+      Key
+      trackKey (PAR parent, DEF specialisation)
+        {
+          Key newKey (parent,specialisation);
+          maybeStore (newKey);
+          return newKey;
+        }
+      
+      void
+      maybeStore (Key const& key)
+        {
+          if (isKnown (key)) return;
+          UNIMPLEMENTED ("registry for type keys");
+        }
+
+    };
+    
+    
+    
+    
+  
+  
+  
+  
+  /** */
+  inline Metadata::Entry&
+  Metadata::markLocked (Key const& parentKey, const void* buffer)
+  {
+    UNIMPLEMENTED ("transition to locked state");
+    if (!buffer)
+      throw error::Fatal ("Attempt to lock for a NULL buffer. Allocation floundered?"
+                         , error::LUMIERA_ERROR_BOTTOM_VALUE);
+    
+    Key newKey = this->key (parentKey, buffer);
+    if (isLocked(newKey))
+      throw error::Logic ("Attempt to lock a slot for a new buffer, "
+                          "while actually the old buffer is still locked."
+                         , error::LUMIERA_ERROR_LIFECYCLE );
+    
+    return this->get(newKey);
+  }
+  
+  inline void
+  Metadata::release (HashVal key)
+  {
+    UNIMPLEMENTED ("metadata memory management");
+  }
+  
+  
+  
+} // namespace engine
+#endif

src/proc/engine/buffer-provider.cpp

@@ -0,0 +1,82 @@
+/*
+  BufferProvider  -  Abstraction for Buffer management during playback/render
+
+  Copyright (C)         Lumiera.org
+    2008,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+* *****************************************************/
+
+
+#include "proc/engine/buffer-provider.hpp"
+#include "proc/engine/buffer-metadata.hpp"
+
+
+namespace engine {
+  
+  
+  namespace { // impl. details and definitions
+    
+    const uint DEFAULT_DESCRIPTOR = 0;
+    
+  }
+
+  
+  BufferProvider::BufferProvider (Literal implementationID)
+    : meta_(new Metadata (implementationID))
+    { }
+  
+  BufferProvider::~BufferProvider() { }
+  
+  
+  /** @internal verify the given descriptor.
+   *  @return true if it corresponds to a buffer
+   *          currently locked and usable by client code
+   */
+  bool
+  BufferProvider::verifyValidity (BufferDescriptor const&)
+  {
+    UNIMPLEMENTED ("BufferProvider basic and default implementation");
+  }
+  
+  
+  BufferDescriptor
+  BufferProvider::getDescriptorFor (size_t storageSize)
+  {
+    return BufferDescriptor (*this, meta_->key (storageSize));
+  }
+      
+      
+  
+  
+  /* === BufferDescriptor and BuffHandle === */
+  
+  bool
+  BufferDescriptor::verifyValidity()  const
+  {
+    return provider_->verifyValidity(*this);
+  }
+  
+  
+  void
+  BuffHandle::release()
+  {
+    UNIMPLEMENTED ("forward buffer release call to buffer provider");
+  }
+  
+
+
+} // namespace engine

src/proc/engine/buffer-provider.hpp

@@ -0,0 +1,135 @@
+/*
+  BUFFER-PROVIDER.hpp  -  Abstraction for Buffer management during playback/render
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+/** @file buffer-provider.hpp
+ ** Abstraction to represent buffer management and lifecycle within the render engine.
+ ** It turns out that --  throughout the render engine implementation -- we never need
+ ** direct access to the buffers holding media data. Buffers are just some entity to be \em managed,
+ ** i.e. "allocated", "locked" and "released"; the actual meaning of these operations is an implementation detail.
+ ** The code within the render engine just pushes around BufferHandle objects, which act as a front-end,
+ ** being created by and linked to a BufferProvider implementation. There is no need to manage the lifecycle
+ ** of buffers automatically, because the use of buffers is embedded into the render calculation cycle,
+ ** which follows a rather strict protocol anyway. Relying on the capabilities of the scheduler,
+ ** the sequence of individual jobs in the engine ensures...
+ ** - that the availability of a buffer was ensured prior to planning a job ("buffer allocation")
+ ** - that a buffer handle was obtained ("locked") prior to any operation requiring a buffer
+ ** - that buffers are marked as free ("released") after doing the actual calculations.
+ ** 
+ ** @see state.hpp
+ ** @see output-slot.hpp
+ */
+
+#ifndef PROC_ENGINE_BUFFR_PROVIDER_H
+#define PROC_ENGINE_BUFFR_PROVIDER_H
+
+
+#include "lib/error.hpp"
+#include "lib/symbol.hpp"
+#include "proc/engine/buffhandle.hpp"
+
+#include <boost/noncopyable.hpp>
+#include <boost/scoped_ptr.hpp>
+
+
+namespace engine {
+  
+  using boost::scoped_ptr;
+  using lib::Literal;
+  
+  
+  class Metadata;
+  
+  
+  /**
+   * Interface: a facility providing and managing working buffers for media calculations.
+   * The pointer to actual buffer storage can be retrieved by
+   * - optionally announcing the required buffer(s) beforehand
+   * - "locking" a buffer to yield a buffer handle
+   * - dereferencing this smart-handle class
+   * 
+   * @todo as of 6/2011 buffer management within the engine is still a bit vague
+   */
+  class BufferProvider
+    : boost::noncopyable
+    {
+      scoped_ptr<Metadata> meta_;
+      
+    protected:
+      BufferProvider (Literal implementationID);
+      
+    public:
+      virtual ~BufferProvider();  ///< this is an interface
+      
+      
+      virtual uint announce (uint count, BufferDescriptor const&) =0;
+      
+      virtual BuffHandle lockBufferFor (BufferDescriptor const&)  =0;
+      virtual void releaseBuffer (BuffHandle const&)              =0;
+      
+      template<typename BU>
+      BuffHandle lockBufferFor ();
+      
+      
+      /** describe the kind of buffer managed by this provider */
+      BufferDescriptor getDescriptorFor(size_t storageSize=0);
+      
+      template<typename BU>
+      BufferDescriptor getDescriptor();
+
+      
+      
+      /* === API for BuffHandle internal access === */
+      
+      bool verifyValidity (BufferDescriptor const&);
+      
+    };
+    
+    
+    
+    
+  /* === Implementation === */
+  
+  /** convenience shortcut:
+   *  prepare and claim ("lock") a buffer suitable
+   *  to hold an object of the given type.
+   * @return a handle embedding a suitably configured
+   *         buffer descriptor. The corresponding buffer
+   *         has been allocated and marked for exclusive use
+   */
+  template<typename BU>
+  BuffHandle
+  BufferProvider::lockBufferFor()
+  {
+    UNIMPLEMENTED ("convenience shortcut to announce and lock for a specific object type");
+  }
+  
+  
+  template<typename BU>
+  BufferDescriptor
+  BufferProvider::getDescriptor()
+  {
+    UNIMPLEMENTED ("build descriptor for automatically placing an object instance into the buffer");
+  }
+  
+  
+} // namespace engine
+#endif

src/proc/engine/buffhandle.hpp

@@ -24,12 +24,17 @@
  ** Various bits needed to support the buffer management within the render nodes.
  ** When pulling data from predecessor nodes and calculating new data, each render node
  ** needs several input and output buffers. These may be allocated and provided by several
- ** different "buffer providers" (for example the frame cache). For accessing those buffers,
- ** the node needs to keep a table of buffer pointers, and for releasing the buffers later
- ** on, we need some handles. The usage pattern of those buffer pointer tables is stack-like,
- ** thus it makes sense to utilise a single large buffer pointer array per pull() calldown
- ** sequence and dynamically claim small chunks for each node.
- **
+ ** different "buffer providers" (for example the frame cache). Typically, the real buffers
+ ** will be passed as parameters to the actual job instance when scheduled, drawing on the
+ ** results of prerequisite jobs. Yet the actual job implementation remains agnostic with
+ ** respect to the way actual buffers are provided; the invocation just pushes BuffHandle
+ ** objects around. The actual render function gets an array of C-pointers to the actual
+ ** buffers, and for accessing those buffers, the node needs to keep a table of buffer
+ ** pointers, and for releasing the buffers later on, we utilise the buffer handles.
+ ** The usage pattern of those buffer pointer tables is stack-like, thus the actual
+ ** implementation utilises a single large buffer pointer array per pull() call
+ ** sequence and dynamically claims small chunks for each node.
+ ** 
  ** @see nodewiring-def.hpp
  ** @see nodeoperation.hpp
  ** @see bufftable.hpp       storage for the buffer table
@@ -47,67 +52,48 @@
 
 namespace engine {
   
+  class BufferProvider;
+  
   
   
   /**
-   * Handle for a buffer for processing data, abstracting away the actual implementation.
-   * The real buffer pointer can be retrieved by dereferencing this smart-handle class.
+   * An opaque descriptor to identify the type and further properties of a data buffer.
+   * For each kind of buffer, there is somewhere a BufferProvider responsible for the
+   * actual storage management. This provider may "lock" a buffer for actual use,
+   * returning a BuffHandle.
+   * @note this descriptor and especially the #subClassification_ is really owned
+   *       by the BufferProvider, which may use (and even change) the opaque contents
+   *       to organise the internal buffer management.
+   * 
+   * @todo try to move that definition into buffer-provider.hpp   ////////////////////////////////////TICKET #249
    */
-  struct BuffHandle
-    : lib::BoolCheckable<BuffHandle>
+  class BufferDescriptor
     {
-      typedef lumiera::StreamType::ImplFacade::DataBuffer Buff;
-      typedef Buff* PBuff;
+      BufferProvider* provider_;
+      uint64_t subClassification_;
       
-      PBuff 
-      operator->()  const 
-        { 
-          return pBuffer_; 
-        }
-      Buff&
-      operator* ()  const
-        {
-          ENSURE (pBuffer_);
-          return *pBuffer_;
-        }
+      BufferDescriptor(BufferProvider& manager, uint64_t detail)
+        : provider_(&manager)
+        , subClassification_(detail)
+      { }
       
-      bool
-      isValid()  const
-        {
-          return pBuffer_;
-        }
+      friend class BufferProvider;
       
+    public:
+      // using standard copy operations
       
-      //////////////////////TODO: the whole logic how to create a BuffHandle needs to be solved in a more clever way. --> Ticket 249
-      BuffHandle()
-        : pBuffer_(0),
-          sourceID_(0)
-        { }
-      
-    private:
-      PBuff pBuffer_; 
-      long sourceID_;
+      bool verifyValidity()  const;
     };
   
   
-  /**
-   * Buffer Type information.
-   * Given a BufferDescriptor, it is possible to allocate a buffer
-   * of suitable size and type by using State::allocateBuffer().
-   */
-  struct BufferDescriptor
-    {
-      lumiera::StreamType& sType_;
-    };
-  
   
   class ProcNode;
   typedef ProcNode* PNode;
   
   
-  struct ChannelDescriptor  ///////TODO collapse this with BufferDescriptor?
+  struct ChannelDescriptor  ///////TODO really need to define that here? it is needed for node wiring only
     {
-      BufferDescriptor bufferType;
+      const lumiera::StreamType * bufferType;                /////////////////////////////////////////TICKET #828
     };
   
   struct InChanDescriptor : ChannelDescriptor
@@ -118,5 +104,99 @@ namespace engine {
   
   
   
+  
+  
+  /**
+   * Handle for a buffer for processing data, abstracting away the actual implementation.
+   * The real buffer pointer can be retrieved by dereferencing this smart-handle class.
+   * 
+   * @todo as of 6/2011 it isn't clear how buffer handles are actually created
+   *       and how the lifecycle (and memory) management works                  //////////////////////TICKET #249 rework BuffHandle creation and usage
+   */
+  class BuffHandle
+    : public lib::BoolCheckable<BuffHandle>
+    {
+      typedef lumiera::StreamType::ImplFacade::DataBuffer Buff;
+      
+      BufferDescriptor descriptor_;
+      Buff* pBuffer_; 
+      
+      
+    public:
+      typedef Buff* PBuff;
+      
+      /** @internal a buffer handle may be obtained by "locking"
+       *  a buffer from the corresponding BufferProvider */
+      BuffHandle(BufferDescriptor const& typeInfo, PBuff storage = 0)
+        : descriptor_(typeInfo)
+        , pBuffer_(storage)
+        { }
+      
+      // using standard copy operations
+      
+      
+      
+      void release();
+      
+      
+      template<typename BU>
+      BU& create();
+      
+      template<typename BU>
+      BU& accessAs();
+      
+      
+      Buff&
+      operator* ()  const
+        {
+          ENSURE (pBuffer_);
+          return *pBuffer_;
+        }
+      
+      bool
+      isValid()  const
+        {
+          return bool(pBuffer_)
+              && descriptor_.verifyValidity();
+        }
+      
+      size_t
+      size()  const
+        {
+          UNIMPLEMENTED ("forward to the buffer provider for storage size diagnostics");
+        }
+      
+    };
+  
+  
+  /* === Implementation details === */
+  
+  /** convenience shortcut: place and maintain an object within the buffer.
+   *  This operation performs the necessary steps to attach an object;
+   *  if the buffer isn't locked yet, it will do so. Moreover, the created
+   *  object will be owned by the buffer management facilities, i.e. the
+   *  destructor is registered as cleanup function.   
+   */
+  template<typename BU>
+  BU&
+  BuffHandle::create()
+  {
+    UNIMPLEMENTED ("convenience shortcut to attach/place an object in one sway");
+  }
+  
+  
+  /** convenience shortcut: access the buffer contents in a typesafe fashion.
+   *  This is equivalent to a plain dereferentiation with additional metadata check
+   * @throw error::Logic in case of type mismatch \c LUMIERA_ERROR_WRONG_TYPE
+   */
+  template<typename BU>
+  BU&
+  BuffHandle::accessAs()
+  {
+    UNIMPLEMENTED ("convenience shortcut to access buffer contents typesafe");
+  }
+  
+  
+  
 } // namespace engine
 #endif

src/proc/engine/bufftable-obsolete.hpp

@@ -0,0 +1,186 @@
+/*
+  BUFFTABLE-OBSOLTE.hpp  -  Old dead code to be removed when rewriting ProcNode!!!!!
+
+  Copyright (C)         Lumiera.org
+    2008,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+#ifndef ENGINE_BUFFHTABLE_OBSOLETE_H
+#define ENGINE_BUFFHTABLE_OBSOLETE_H
+
+
+#include "lib/error.hpp"
+#include "proc/engine/buffhandle.hpp"
+#include "proc/engine/procnode.hpp"
+
+#include <boost/noncopyable.hpp>
+#include <vector>
+#include <utility>
+
+
+////////////////////////////////WARNING: obsolete code
+////////////////////////////////WARNING: ...just left in tree to keep it compiling
+////////////////////////////////TICKET   #826  need to be reworked entirely
+
+namespace engine {
+  
+  using std::pair;
+  using std::vector;
+  
+  
+    /**
+     * Obsolete, to be rewritten  /////TICKET #826
+     *  
+     * Tables of buffer handles and corresponding dereferenced buffer pointers.
+     * Used within the invocation of a processing node to calculate data.
+     * The tables are further differentiated into input data buffers and output
+     * data buffers. The tables are supposed to be implemented as bare "C" arrays,
+     * thus the array of real buffer pointers can be fed directly to the
+     * processing function of the respective node.
+     * 
+     * @todo this whole design is a first attempt and rather clumsy. It should be reworked
+     *       to use a single contiguous memory area and just layer the object structure on top
+     *       (by using placement new). Yet the idea of an stack-like organisation should be retained
+     */
+  struct BuffTable
+    {
+      typedef BuffHandle        * PHa;
+      typedef BuffHandle::PBuff * PBu;
+      typedef pair<PHa const,PBu const> Chunk;
+      
+      PHa outHandle;
+      PHa inHandle;
+      PBu outBuff;
+      PBu inBuff;
+    };
+  
+  
+    /** Obsolete, to be rewritten  /////TICKET #826 */
+  class BuffTableStorage
+    {
+      /////////////////////////////////////////////////////////////////////////TICKET #826  need to be reworked entirely
+      /** just a placeholder to decouple the existing code
+       *  from the reworked BuffHandle logic. The existing
+       *  code in turn will be reworked rather fundamentally
+       */
+      struct BuffHaXXXX
+        : BuffHandle
+        {
+          BuffHaXXXX() : BuffHandle(just_satisfy_the_compiler()) { /* wont work ever */ }
+          static BufferDescriptor const& 
+          just_satisfy_the_compiler() { }
+        };
+        
+                                           ////////////////////////////////////TICKET #825  should be backed by mpool and integrated with node invocation
+      vector<BuffHaXXXX>        hTab_;
+      vector<BuffHandle::PBuff> pTab_;
+      size_t level_;
+      
+    public:
+      BuffTableStorage (const size_t maxSiz)
+        : hTab_(maxSiz),
+          pTab_(maxSiz),
+          level_(0)
+        { }
+      
+      ~BuffTableStorage() { ASSERT (0==level_, "buffer management logic broken."); }
+      
+    protected:
+      
+      friend class BuffTableChunk;
+      
+      /** allocate the given number of slots
+       *  starting at current level to be used
+       *  by the newly created BuffTableChunk
+       */
+      BuffTable::Chunk
+      claim (uint slots)
+        {
+          ASSERT (pTab_.size() == hTab_.size());
+          REQUIRE (level_+slots <= hTab_.size());
+          
+          size_t prev_level (level_);
+          level_ += slots;
+          return std::make_pair (&hTab_[prev_level],
+                                 &pTab_[prev_level]);
+        }
+      
+      void
+      release (uint slots)
+        {
+          ASSERT (slots <= level_);
+          REQUIRE (level_ <= hTab_.size());
+          REQUIRE (level_ <= pTab_.size());
+          
+          level_ -= slots;
+        }
+      
+      bool
+      level_check (BuffTable::Chunk& prev_level)
+        {
+          return prev_level.first  == &hTab_[level_]
+              && prev_level.second == &pTab_[level_];
+        }
+    };
+  
+  
+  /** Obsolete, to be rewritten  /////TICKET #826 
+   * to be allocated on the stack while evaluating a ProcNode#pull() call.
+   * The "current" State (StateProxy) maintains a BuffTableStorage (=pool),
+   * which can be used to crate such chunks. The claiming and releasing of
+   * slots in the BuffTableStorage is automatically tied to BuffTableChunk
+   * object's lifecycle.
+   */
+  class BuffTableChunk
+    : public BuffTable,
+      boost::noncopyable
+    {
+      const uint siz_;
+      BuffTable::Chunk tab_;
+      BuffTableStorage& sto_;
+      
+    public:
+      BuffTableChunk (WiringDescriptor const& wd, BuffTableStorage& storage)
+        : siz_(wd.nrI + wd.nrO),
+          tab_(storage.claim (siz_)),
+          sto_(storage)
+        {
+          const uint nrO(wd.nrO);
+          
+          // Setup the publicly visible table locations
+          this->outHandle = &tab_.first[ 0 ];
+          this->inHandle  = &tab_.first[nrO];
+          this->outBuff   = &tab_.second[ 0 ];
+          this->inBuff    = &tab_.second[nrO];
+        }
+      
+      ~BuffTableChunk ()
+        {
+          sto_.release (siz_);
+          ASSERT ( sto_.level_check (tab_),
+                  "buffer management logic broken.");
+        }
+    };
+  
+  
+  
+  
+  
+} // namespace engine
+#endif

src/proc/engine/bufftable.hpp

@@ -28,6 +28,7 @@
 #include "lib/error.hpp"
 #include "proc/engine/buffhandle.hpp"
 #include "proc/engine/procnode.hpp"
+#include "lib/iter-adapter.hpp"
 
 #include <boost/noncopyable.hpp>
 #include <vector>
@@ -48,113 +49,108 @@ namespace engine {
      * data buffers. The tables are supposed to be implemented as bare "C" arrays,
      * thus the array of real buffer pointers can be fed directly to the
      * processing function of the respective node.
+     * 
+     * @todo this whole design is a first attempt and rather clumsy. It should be reworked
+     *       to use a single contiguous memory area and just layer the object structure on top
+     *       (by using placement new). Yet the idea of an stack-like organisation should be retained
      */
   struct BuffTable
     {
       typedef BuffHandle        * PHa;
       typedef BuffHandle::PBuff * PBu;
-      typedef pair<PHa const,PBu const> Chunk;
       
-      PHa outHandle;
-      PHa inHandle;
-      PBu outBuff;
-      PBu inBuff;
+      struct StorageChunk
+        { };
+      
+      template<uint count>
+      struct Storage
+        {
+          enum{size = count * sizeof(StorageChunk)};
+        };
+      
+      class Builder
+        {
+        public:
+          Builder& announce (uint count, BufferDescriptor const& type);
+          BuffTable& build();
+        };
+      
+      static Builder& prepare (const size_t STORAGE_SIZE, void* storage);
+      
+      void lockBuffers();
+      void releaseBuffers();
+      
+      typedef vector<BuffHandle> BuffHandleTable;
+      typedef lib::RangeIter<BuffHandleTable::iterator> iterator;
+      
+      iterator buffers();
+      iterator inBuffers();
+      iterator outBuffers();
     };
   
   
-  class BuffTableStorage
-    {
-      vector<BuffHandle>        hTab_;
-      vector<BuffHandle::PBuff> pTab_;
-      size_t level_;
-      
-    public:
-      BuffTableStorage (const size_t maxSiz)
-        : hTab_(maxSiz),
-          pTab_(maxSiz),
-          level_(0)
-        { }
-      
-      ~BuffTableStorage() { ASSERT (0==level_, "buffer management logic broken."); }
-      
-    protected:
-      
-      friend class BuffTableChunk;
-      
-      /** allocate the given number of slots
-       *  starting at current level to be used
-       *  by the newly created BuffTableChunk
-       */
-      BuffTable::Chunk
-      claim (uint slots)
-        {
-          ASSERT (pTab_.size() == hTab_.size());
-          REQUIRE (level_+slots <= hTab_.size());
-          
-          size_t prev_level (level_);
-          level_ += slots;
-          return std::make_pair (&hTab_[prev_level],
-                                 &pTab_[prev_level]);
-        }
-      
-      void
-      release (uint slots)
-        {
-          ASSERT (slots <= level_);
-          REQUIRE (level_ <= hTab_.size());
-          REQUIRE (level_ <= pTab_.size());
-          
-          level_ -= slots;
-        }
-      
-      bool
-      level_check (BuffTable::Chunk& prev_level)
-        {
-          return prev_level.first  == &hTab_[level_]
-              && prev_level.second == &pTab_[level_];
-        }
-    };
   
   
-  /** 
-   * to be allocated on the stack while evaluating a ProcNode#pull() call.
-   * The "current" State (StateProxy) maintains a BuffTableStorage (=pool),
-   * which can be used to crate such chunks. The claiming and releasing of
-   * slots in the BuffTableStorage is automatically tied to BuffTableChunk
-   * object's lifecycle.
-   */
-  class BuffTableChunk
-    : public BuffTable,
-      boost::noncopyable
-    {
-      const uint siz_;
-      BuffTable::Chunk tab_;
-      BuffTableStorage& sto_;
-      
-    public:
-      BuffTableChunk (WiringDescriptor const& wd, BuffTableStorage& storage)
-        : siz_(wd.nrI + wd.nrO),
-          tab_(storage.claim (siz_)),
-          sto_(storage)
-        {
-          const uint nrO(wd.nrO);
-          
-          // Setup the public visible table locations
-          this->outHandle = &tab_.first[ 0 ];
-          this->inHandle  = &tab_.first[nrO];
-          this->outBuff   = &tab_.second[ 0 ];
-          this->inBuff    = &tab_.second[nrO];
-        }
-      
-      ~BuffTableChunk ()
-        {
-          sto_.release (siz_);
-          ASSERT ( sto_.level_check (tab_),
-                  "buffer management logic broken.");
-        }
-    };
+  /* === Implementation === */
+  
+  inline BuffTable::Builder&
+  BuffTable::prepare (const size_t STORAGE_SIZE, void* storage)
+  {
+    UNIMPLEMENTED ("expose a builder object for outfitting a buffer pointer table");
+  }
   
   
+  inline BuffTable::Builder&
+  BuffTable::Builder::announce (uint count, BufferDescriptor const& type)
+  {
+    UNIMPLEMENTED ("accept announcement of additional buffer table entries required");
+  }
+  
+  
+  inline BuffTable&
+  BuffTable::Builder::build()
+  {
+    UNIMPLEMENTED ("finally drop off the newly configured buffer pointer table");
+  }
+  
+  
+  inline void
+  BuffTable::lockBuffers()
+  {
+    UNIMPLEMENTED ("convenience shortcut: lock all preconfigured buffers within this table through the underlying buffer provider");
+  }
+  
+  
+  inline void
+  BuffTable::releaseBuffers()
+  {
+    UNIMPLEMENTED ("convenience shortcut: release all the buffers managed through this buffer table, by forwarding to the underlying buffer provider");
+  }
+  
+  
+  
+  inline BuffTable::iterator
+  BuffTable::buffers()
+  {
+    UNIMPLEMENTED ("expose an iterator to yield all prepared buffers within this buffer table");
+  }
+  
+  
+  inline BuffTable::iterator
+  BuffTable::inBuffers()
+  {
+    UNIMPLEMENTED ("expose an iterator to access all the input buffer slots of this buffer table");
+  }
+  
+  
+  inline BuffTable::iterator
+  BuffTable::outBuffers()
+  {
+    UNIMPLEMENTED ("expose an iterator to access all the output buffer slots of this buffer table");
+  }
+
+
+
   
   
   

src/proc/engine/calc-stream.hpp

@@ -0,0 +1,94 @@
+/*
+  CALC-STREAM.hpp  -  abstraction representing a series of scheduled calculations
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+
+#ifndef PROC_ENGINE_CALC_STREAM_H
+#define PROC_ENGINE_CALC_STREAM_H
+
+
+#include "lib/error.hpp"
+//#include "include/dummy-player-facade.h"
+//#include "include/display-facade.h"
+//#include "common/instancehandle.hpp"
+//#include "lib/singleton-ref.hpp"
+//
+//#include <boost/noncopyable.hpp>
+//#include <boost/scoped_ptr.hpp>
+//#include <string>
+
+
+namespace proc  {
+namespace engine{
+
+//    using std::string;
+//    using lumiera::Subsys;
+//    using lumiera::Display;
+//    using lumiera::DummyPlayer;
+  
+  
+  
+  
+  
+  /*********************************************************
+   * A calculation stream groups and abstracts a series of
+   * calculation jobs, delivering frames into the configured
+   * OutputSlot in a timely fashion. Behind the scenes, this
+   * gets translated into several jobs enqueued with the
+   * scheduler in the backend layer. The calculation stream
+   * implementation cares to create and configure these
+   * jobs and to manage the necessary dependencies and
+   * callbacks.
+   * 
+   * Regarding the implementation, a CalcStream is an const
+   * value object holding the metadata necessary to manage
+   * the underlying jobs. The only way to create a CalcStream
+   * properly is to retrieve it from the  factory functions
+   * of the EngineService. At that point, the corresponding
+   * jobs will be configured and enqueued. 
+   */
+  class CalcStream
+    {
+      
+      friend class EngineService;
+      
+      CalcStream()
+        {
+          
+        }
+      
+    public:
+      CalcStream (CalcStream const& o)
+        { 
+          UNIMPLEMENTED("build a calculation stream");
+        }
+      
+     ~CalcStream() { }
+      
+    };
+  
+  
+  
+  
+} // namespace engine
+} // namespace proc
+#endif

src/proc/engine/diagnostic-buffer-provider.cpp

@@ -0,0 +1,220 @@
+/*
+  DiagnosticBufferProvider  -  helper for testing against the BufferProvider interface
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+* *****************************************************/
+
+
+#include "lib/error.hpp"
+#include "include/logging.h"
+#include "lib/meta/function.hpp"
+#include "lib/scoped-ptrvect.hpp"
+
+#include "proc/engine/diagnostic-buffer-provider.hpp"
+
+#include <boost/scoped_array.hpp>
+//#include <vector>
+
+using lib::ScopedPtrVect;
+using boost::scoped_array;
+
+
+
+namespace engine {
+  
+  
+  /** Storage for the diagnostics frontend */
+  lib::Singleton<DiagnosticBufferProvider> DiagnosticBufferProvider::diagnostics;
+  
+  
+  class Block
+    : boost::noncopyable
+    {
+      size_t size_;
+      scoped_array<char> storage_;
+      
+      bool was_locked_;
+      
+    public:
+      Block()
+        : size_(0)
+        , storage_()
+        , was_locked_(false)
+        { }
+      
+      bool
+      was_used()  const
+        {
+          return was_locked_;
+        }
+      
+      bool
+      was_closed()  const
+        {
+          return was_locked_;
+        }
+      
+      void*
+      accessMemory()  const
+        {
+          return storage_.get();
+        }
+    };
+  
+    
+    
+  namespace { // Details of allocation and accounting
+    
+    const uint MAX_BUFFERS = 50;
+  
+  } // (END) Details of allocation and accounting
+  
+  
+  
+  /**
+   * @internal DiagnosticBufferProvider's PImpl.
+   * Uses a linearly growing table of heap allocated buffer blocks,
+   * which will never be discarded, unless the PImpl is discarded as a whole.
+   * This way, the tracked usage information remains available after the fact.
+   */
+  class DiagnosticBufferProvider::HeapMemProvider
+    : public BufferProvider
+    , public ScopedPtrVect<Block>
+    {
+      
+      virtual uint
+      announce (uint count, BufferDescriptor const& type)
+        {
+          UNIMPLEMENTED ("pre-register storage for buffers of a specific kind");   
+        }
+
+      
+      virtual BuffHandle
+      lockBufferFor (BufferDescriptor const& descriptor)
+        {
+          UNIMPLEMENTED ("lock buffer for exclusive use");
+        }
+      
+      virtual void
+      releaseBuffer (BuffHandle const& handle)
+        {
+          UNIMPLEMENTED ("release a buffer and invalidate the handle");
+        }
+      
+    public:
+      HeapMemProvider()
+        : BufferProvider ("Diagnostic_HeapAllocated")
+        { }
+      
+      virtual ~HeapMemProvider()
+        {
+          INFO (proc_mem, "discarding %zu diagnostic buffer entries", HeapMemProvider::size());
+        }
+      
+      Block&
+      access_or_create (uint bufferID)
+        {
+          while (!withinStorageSize (bufferID))
+            manage (new Block);
+          
+          ENSURE (withinStorageSize (bufferID));
+          return (*this)[bufferID];
+        }
+      
+    private:
+      bool
+      withinStorageSize (uint bufferID)  const
+        {
+          if (bufferID >= MAX_BUFFERS)
+            throw error::Fatal ("hardwired internal limit for test buffers exceeded");
+          
+          return bufferID < size();
+        }
+    };
+  
+  
+
+  DiagnosticBufferProvider::DiagnosticBufferProvider()
+    : pImpl_() //////////TODO create PImpl here
+    { }
+  
+  DiagnosticBufferProvider::~DiagnosticBufferProvider() { }
+  
+  
+  BufferProvider&
+  DiagnosticBufferProvider::build()
+  {
+    return diagnostics().reset();
+  }
+  
+  
+  DiagnosticBufferProvider&
+  DiagnosticBufferProvider::access (BufferProvider const& provider)
+  {
+    if (!diagnostics().isCurrent (provider))
+      throw error::Invalid("given Provider doesn't match (current) diagnostic data record."
+                           "This might be an lifecycle error. Did you build() this instance beforehand?");
+    
+    return diagnostics();
+  }
+  
+
+    
+    
+  DiagnosticBufferProvider::HeapMemProvider&
+  DiagnosticBufferProvider::reset()
+  {
+    pImpl_.reset(new HeapMemProvider());
+    return *pImpl_;
+  }
+  
+  bool
+  DiagnosticBufferProvider::isCurrent (BufferProvider const& implInstance)
+  {
+    return &implInstance == pImpl_.get();
+  }
+
+  
+
+  
+  
+  /* === diagnostic API === */
+  
+  bool
+  DiagnosticBufferProvider::buffer_was_used (uint bufferID)  const
+    {
+      return pImpl_->access_or_create(bufferID).was_used();
+    }
+  
+  
+  bool
+  DiagnosticBufferProvider::buffer_was_closed (uint bufferID)  const
+    {
+      return pImpl_->access_or_create(bufferID).was_closed();
+    }
+  
+  
+  void*
+  DiagnosticBufferProvider::accessMemory (uint bufferID)  const
+    {
+      return pImpl_->access_or_create(bufferID).accessMemory();
+    }
+  
+
+} // namespace engine

src/proc/engine/diagnostic-buffer-provider.hpp

@@ -0,0 +1,125 @@
+/*
+  DIAGNOSTIC-BUFFER-PROVIDER.hpp  -  helper for testing against the BufferProvider interface
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+/** @file diagnostic-buffer-provider.hpp
+ ** An facility for writing unit-tests targetting the BufferProvider interface.
+ **
+ ** @see buffer-provider-protocol-test.cpp
+ */
+
+#ifndef PROC_ENGINE_DIAGNOSTIC_BUFFR_PROVIDER_H
+#define PROC_ENGINE_DIAGNOSTIC_BUFFR_PROVIDER_H
+
+
+#include "lib/error.hpp"
+#include "lib/singleton.hpp"
+#include "lib/util.hpp"
+#include "proc/engine/buffer-provider.hpp"
+
+#include <boost/scoped_ptr.hpp>
+#include <boost/noncopyable.hpp>
+
+
+namespace engine {
+  
+  namespace error = lumiera::error;
+  
+  
+  /********************************************************************
+   * Helper for unit tests: Buffer provider reference implementation.
+   * 
+   * @todo write type comment
+   */
+  class DiagnosticBufferProvider
+    : boost::noncopyable
+    {
+      
+      /**
+       * simple BufferProvider implementation
+       * with additional allocation tracking
+       */
+      class HeapMemProvider;
+      
+      
+      boost::scoped_ptr<HeapMemProvider>              pImpl_;
+      static lib::Singleton<DiagnosticBufferProvider> diagnostics;
+      
+      
+      HeapMemProvider& reset();
+      bool isCurrent (BufferProvider const&);
+      
+      
+      DiagnosticBufferProvider();
+     ~DiagnosticBufferProvider();
+     
+      friend class lib::singleton::StaticCreate<DiagnosticBufferProvider>;
+     
+    public:
+      /** build a new Diagnostic Buffer Provider instance,
+       *  discard the existing one. Use the static query API
+       *  for investigating collected data. */
+      static BufferProvider& build();
+      
+      
+      /** access the diagnostic API of the buffer provider
+       * @throw error::Invalid if the given provider doesn't allow
+       *        for diagnostic access or wasn't registered beforehand.
+       */
+      static DiagnosticBufferProvider&
+      access (BufferProvider const& );
+      
+      
+      
+      
+      /* === diagnostic API === */
+      
+      bool buffer_was_used (uint bufferID)  const;
+      bool buffer_was_closed (uint bufferID) const;
+      void* accessMemory (uint bufferID)   const;
+      bool all_buffers_released()          const;
+      
+      
+      template<typename BU>
+      bool
+      object_was_attached (uint bufferID)  const
+        {
+          UNIMPLEMENTED ("verify object attachment status of a specific buffer");
+        }
+      
+      
+      template<typename BU>
+      bool
+      object_was_destroyed (uint bufferID)  const
+        {
+          UNIMPLEMENTED ("verify object attachment status of a specific buffer");
+        }
+      
+      
+      
+    private:
+      
+    };
+  
+  
+  
+} // namespace engine
+#endif

src/proc/engine/dispatcher.cpp

@@ -1,8 +1,8 @@
 /*
-  BuffHandle  -  Buffer handling support for the render engine
+  Dispatcher  -  translating calculation streams into frame jobs
 
   Copyright (C)         Lumiera.org
-    2008,               Hermann Vosseler <Ichthyostega@web.de>
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
 
   This program is free software; you can redistribute it and/or
   modify it under the terms of the GNU General Public License as
@@ -21,12 +21,17 @@
 * *****************************************************/
 
 
-#include "proc/engine/buffhandle.hpp"
+#include "proc/engine/dispatcher.hpp"
+//#include "lib/frameid.hpp"
+//#include "proc/state.hpp"
+
+
 
 namespace engine {
-
-
-  /**  */
-
-
+  
+  /** */
+  
+  
+  
+  
 } // namespace engine

src/proc/engine/dispatcher.hpp

@@ -0,0 +1,62 @@
+/*
+  DISPATCHER.hpp  -  translating calculation streams into frame jobs
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+
+#ifndef PROC_ENGINE_DISPATCHER_H
+#define PROC_ENGINE_DISPATCHER_H
+
+#include "proc/common.hpp"
+#include "proc/state.hpp"
+#include "lib/time/timevalue.hpp"
+
+
+
+namespace engine {
+  
+  using lib::time::TimeSpan;
+  using lib::time::FSecs;
+  using lib::time::Time;
+//  
+//  class ExitNode;
+  
+  /**
+   * @todo
+   */
+  class Dispatcher
+    {
+    protected:
+      /** timerange covered by this RenderGraph */
+      TimeSpan segment_;
+      
+    public:
+      Dispatcher()
+        : segment_(Time::ZERO, FSecs(5))
+        {
+          UNIMPLEMENTED ("anything regarding the Engine backbone");
+        }
+      
+    };
+  
+  
+  
+} // namespace engine
+#endif

src/proc/engine/engine-diagnostics.hpp

@@ -0,0 +1,113 @@
+/*
+  ENGINE-DIAGNOSTICS.hpp  -  diagnostic facility to investigate engine operation
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+/** @file engine-diagnostics.hpp
+ ** An facility to check and monitor engine operations.
+ ** Once created, an EngineDiagnostics object connects to the EngineService
+ ** interface to activate additional tracing facilities within the engine,
+ ** allowing to watch and verify the creation of individual jobs and further
+ ** engine state parameters.
+ ** 
+ ** @see EngineInterface
+ ** @see engine-interface-test.cpp
+ ** @see calc-stream-test.cpp
+ */
+
+
+#ifndef PROC_ENGINE_ENGINE_DIAGNOSTICS_H
+#define PROC_ENGINE_ENGINE_DIAGNOSTICS_H
+
+
+#include "lib/error.hpp"
+#include "proc/engine/engine-service.hpp"
+//#include "include/dummy-player-facade.h"
+//#include "include/display-facade.h"
+//#include "proc/engine/calc-stream.hpp"
+//#include "proc/mobject/model-port.hpp"
+#include "proc/play/timings.hpp"
+//#include "proc/play/output-slot.hpp"
+//#include "common/instancehandle.hpp"
+//#include "lib/singleton-ref.hpp"
+//#include "lib/polymorphic-value.hpp"
+//#include "lib/singleton.hpp"
+//
+#include <boost/noncopyable.hpp>
+//#include <boost/scoped_ptr.hpp>
+//#include <string>
+
+
+namespace proc  {
+namespace engine{
+
+//    using std::string;
+//    using lumiera::Subsys;
+//    using lumiera::Display;
+//    using lumiera::DummyPlayer;
+  using proc::play::Timings;
+  
+  
+  
+  
+  
+  /******************************************************
+   * Render engine diagnostic facility. Creating an instance
+   * will activate additional tracing facilities within the
+   * render engine; results may be investigated through
+   * EngineDiagnostics public functions. The object acts
+   * like a smart handle, i.e. the tracing facilities will
+   * be disabled and disconnected when going out of scope.
+   */
+  class EngineDiagnostics
+    : boost::noncopyable
+    {
+      EngineService& engine_;
+      
+    public:
+      EngineDiagnostics (EngineService& e)
+        : engine_(e)
+        {
+          UNIMPLEMENTED ("attach tracing connector");
+          engine_.activateTracing();
+        }
+      
+     ~EngineDiagnostics()
+        {
+          TODO ("detach tracing connector");
+          engine_.disableTracing();
+        }
+      
+      /** */
+      bool
+      has_scheduled_jobs_for (Timings const& timings)
+        {
+          UNIMPLEMENTED ("Engine Diagnostics: query scheduled jobs");
+        }
+    };
+  
+  
+  
+  
+  
+  
+} // namespace engine
+} // namespace proc
+#endif

src/proc/engine/engine-service.cpp

@@ -0,0 +1,135 @@
+/*
+  EngineService  -  primary service access point for using the renderengine
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+* *****************************************************/
+
+
+#include "proc/engine/engine-service.hpp"
+
+//#include <string>
+//#include <memory>
+//#include <tr1/functional>
+//#include <boost/scoped_ptr.hpp>
+
+
+
+namespace proc  {
+namespace engine{
+  
+//    using std::string;
+//    using lumiera::Subsys;
+//    using std::auto_ptr;
+//    using boost::scoped_ptr;
+//    using std::tr1::bind;
+  
+  
+  namespace { // hidden local details of the service implementation....
+    
+  } // (End) hidden service impl details
+  
+  
+  
+  /** storage for the EngineService interface object */
+  lib::Singleton<EngineService> EngineService::instance;
+  
+  
+  
+  /** */  
+  EngineService::EngineService()
+    { }
+  
+  
+  
+  /** */
+  CalcStream
+  EngineService::calculate(ModelPort mPort,
+                           Timings nominalTimings,
+                           OutputConnection output,
+                           Quality serviceQuality)
+  {
+    UNIMPLEMENTED ("build a standard calculation stream");
+  }
+  
+  
+  
+  /** */
+  CalcStream
+  EngineService::calculateBackground(ModelPort mPort,
+                                     Timings nominalTimings,
+                                     Quality serviceQuality)
+  {
+    UNIMPLEMENTED ("build a calculation stream for background rendering");
+  }
+  
+  
+  
+  /* ===== Quality-of-Service ===== */
+  
+  
+  EngineService::Quality::~Quality() { } // emit vtables here...
+  
+  enum CalcType {
+    PLAYBACK,
+    RENDER,
+    BACKGROUND
+  };
+  
+  
+  class DefaultQoS
+    : public EngineService::Quality
+    {
+      CalcType type_;
+      
+    public:
+      DefaultQoS (CalcType type)
+        : type_(type)
+        { }
+    };
+  
+  class PriorityQoS
+    : public DefaultQoS
+    {
+    
+    public:
+      PriorityQoS ()
+        : DefaultQoS(PLAYBACK)
+        { }
+    };
+  
+  class Compromise
+    : public DefaultQoS
+    {
+      
+    public:
+      Compromise (CalcType type)
+        : DefaultQoS(type)
+        { }
+    };
+  
+  
+  EngineService::QoS_Definition  EngineService::QoS_DEFAULT         = QoS_Definition::build<DefaultQoS> (PLAYBACK);
+  EngineService::QoS_Definition  EngineService::QoS_BACKGROUND      = QoS_Definition::build<DefaultQoS> (BACKGROUND);
+  EngineService::QoS_Definition  EngineService::QoS_COMPROMISE      = QoS_Definition::build<Compromise> (PLAYBACK); 
+  EngineService::QoS_Definition  EngineService::QoS_PERFECT_RESULT  = QoS_Definition::build<DefaultQoS> (RENDER);
+  EngineService::QoS_Definition  EngineService::QoS_SYNC_PRIORITY   = QoS_Definition::build<PriorityQoS>();
+  
+  
+  
+}} // namespace proc::engine

src/proc/engine/engine-service.hpp

@@ -0,0 +1,167 @@
+/*
+  ENGINE-SERVICE.hpp  -  primary service access point for using the renderengine
+
+  Copyright (C)         Lumiera.org
+    2011,               Hermann Vosseler <Ichthyostega@web.de>
+
+  This program is free software; you can redistribute it and/or
+  modify it under the terms of the GNU General Public License as
+  published by the Free Software Foundation; either version 2 of
+  the License, or (at your option) any later version.
+
+  This program is distributed in the hope that it will be useful,
+  but WITHOUT ANY WARRANTY; without even the implied warranty of
+  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+  GNU General Public License for more details.
+
+  You should have received a copy of the GNU General Public License
+  along with this program; if not, write to the Free Software
+  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+*/
+
+/** @file engine-service.hpp
+ ** Access point for the (core) calculation service of the render engine.
+ ** This Proc-Layer internal service is provided for use by the Player subsystem.
+ ** The actual implementation is forwarded to backend services (especially the scheduler).
+ ** The EngineService singleton has no state beyond the jobs currently managed by the
+ ** scheduler; when the latter isn't available, any invocation will throw.
+ ** 
+ ** The central concept provided through this facade interface is the <i>calculation stream</i>.
+ ** This represents a series of calculations, expected to happen in a timely fashion and in order
+ ** to deliver a frame data stream onto an opened output connection. On the implementation side,
+ ** a calculation stream will be translated into a series of jobs invoking render nodes,
+ ** to be executed through the scheduler in the backend layer.
+ ** 
+ ** While the individual CalcStram is simple, linear and unmodifiable, any CalcStream may be
+ ** \em superseded by a new definition. In this case, the engine will care for a seamless
+ ** switch and continuation; under the hood, there is a mechanism to discard resources
+ ** tied to the original CalcStream, once the switch to the new definition is complete.
+ ** 
+ ** @see EngineInterface_test
+ ** @see CalcStream_test
+ ** @see proc::play::PlayerService
+ */
+
+
+#ifndef PROC_ENGINE_ENGINE_SERVICE_H
+#define PROC_ENGINE_ENGINE_SERVICE_H
+
+
+#include "lib/error.hpp"
+//#include "include/dummy-player-facade.h"
+//#include "include/display-facade.h"
+#include "proc/engine/calc-stream.hpp"
+#include "proc/mobject/model-port.hpp"
+#include "proc/play/timings.hpp"
+#include "proc/play/output-slot.hpp"
+//#include "common/instancehandle.hpp"
+//#include "lib/singleton-ref.hpp"
+#include "lib/polymorphic-value.hpp"
+#include "lib/singleton.hpp"
+//
+#include <boost/noncopyable.hpp>
+//#include <boost/scoped_ptr.hpp>
+//#include <string>
+
+
+namespace proc  {
+namespace engine{
+
+//    using std::string;
+//    using lumiera::Subsys;
+//    using lumiera::Display;
+//    using lumiera::DummyPlayer;
+  using mobject::ModelPort;
+  using proc::play::Timings;
+  
+  typedef proc::play::OutputSlot::Allocation OutputConnection;
+  
+  
+  
+  
+  
+  /******************************************************
+   * A service to schedule series of calculations,
+   * delivering the rendered data into an external output
+   * sink in a timely fashion. Actually the CalculationStream
+   * instances provided through this (facade) interface are
+   * backed by jobs executed through the scheduler in the
+   * backend layer. The implementation of this service
+   * cares to create the right job entries in the correct
+   * order and to enqueue these into the scheduler.
+   */
+  class EngineService
+    : boost::noncopyable
+    {
+      
+      
+      /* The following typedefs allow to hand out predefined
+       * Quality-of-Service strategy definitions as value objects,
+       * without disclosing implementation details here in this header.
+       */
+      enum{ QoS_IMPL_SIZE = sizeof(size_t) };  /////////////////////////////////////////TODO is this correct??
+      
+      
+    public:
+      /*************************************************************
+       * Quality-of-Service definition for an Render Engine usage.
+       * This strategy defines how to decide between conflicting goals like
+       * - timely delivery
+       * - image quality
+       * - niceness and resource usage
+       */
+      class Quality
+        {
+          public:
+            virtual ~Quality();  ///< this is an Interface
+        };
+      
+      
+      typedef lib::polyvalue::CloneValueSupport<Quality> _Clonable_QoS_Strategy;
+      typedef lib::PolymorphicValue<Quality, QoS_IMPL_SIZE, _Clonable_QoS_Strategy> QoS_Definition;
+      
+      static QoS_Definition  QoS_DEFAULT;
+      static QoS_Definition  QoS_BACKGROUND;
+      static QoS_Definition  QoS_SYNC_PRIORITY;
+      static QoS_Definition  QoS_PERFECT_RESULT;
+      static QoS_Definition  QoS_COMPROMISE;
+      
+      
+      /** access point to the Engine Interface.
+       * @internal this is an facade interface for internal use
+       *           by the player. Client code should use the Player.
+       */
+      static lib::Singleton<EngineService> instance;
+      
+      
+      EngineService();
+     ~EngineService() { }
+      
+      CalcStream
+      calculate(ModelPort mPort,
+                Timings nominalTimings,
+                OutputConnection output,
+                Quality serviceQuality =QoS_DEFAULT);
+      
+      CalcStream
+      calculateBackground(ModelPort mPort,
+                          Timings nominalTimings,
+                          Quality serviceQuality =QoS_BACKGROUND);
+      
+      
+    protected:
+      void activateTracing();
+      void disableTracing(); ///< EX_FREE
+      
+      friend class EngineDiagnostics;
+    };
+  
+  
+  
+  
+  
+  
+} // namespace engine
+} // namespace proc
+#endif

src/proc/engine/mask.hpp

@@ -24,7 +24,7 @@
 #ifndef ENGINE_MASK_H
 #define ENGINE_MASK_H
 
-#include "proc/engine/trafo.hpp"
+#include "proc/engine/procnode.hpp"
 
 
 
@@ -32,7 +32,7 @@ namespace engine
   {
 
 
-  class Mask : public Trafo
+  class Mask : public ProcNode
     {};
 
 

src/proc/engine/nodeinvocation.hpp

@@ -56,7 +56,7 @@
 #include "proc/state.hpp"
 #include "proc/engine/procnode.hpp"
 #include "proc/engine/buffhandle.hpp"
-#include "proc/engine/bufftable.hpp"
+#include "proc/engine/bufftable-obsolete.hpp"
 
 
 
@@ -172,6 +172,7 @@ namespace engine {
     };
     
     
+                                                                                                    ////////////TICKET #249  this strategy should better be hidden within the BuffHandle ctor (and type-erased after creation)
     struct AllocBufferFromParent  ///< using the parent StateAdapter for buffer allocations
       : Invocation
       {
@@ -179,7 +180,7 @@ namespace engine {
           : Invocation(sta, w, outCh) {}
         
         virtual BuffHandle
-        allocateBuffer (BufferDescriptor const& bd) { return parent_.allocateBuffer(bd); }          ////////////TODO: actually implement the "allocate from parent" logic!
+        allocateBuffer (const lumiera::StreamType* ty) { return parent_.allocateBuffer(ty); }          ////////////TODO: actually implement the "allocate from parent" logic!
       };
     
     struct AllocBufferFromCache   ///< using the global current State, which will delegate to Cache
@@ -189,7 +190,7 @@ namespace engine {
           : Invocation(sta, w, outCh) {}
         
         virtual BuffHandle
-        allocateBuffer (BufferDescriptor const& bd) { return current_.allocateBuffer(bd); }
+        allocateBuffer (const lumiera::StreamType* ty) { return current_.allocateBuffer(ty); }
       };
     
   

src/proc/engine/nodeoperation.hpp

@@ -60,7 +60,7 @@
 #include "proc/state.hpp"
 #include "proc/engine/procnode.hpp"
 #include "proc/engine/buffhandle.hpp"
-#include "proc/engine/bufftable.hpp"
+#include "proc/engine/bufftable-obsolete.hpp"
 #include "proc/engine/nodeinvocation.hpp"
 
 #include "lib/meta/util.hpp"
@@ -245,7 +245,7 @@ namespace config {
   
   template<class NEXT>
   struct ReleaseBuffers : NEXT                     /////////////////TODO: couldn't this be done automatically by BuffTab's dtor??
-    {                                             /////////////////       this would require BuffHandle to be a smart ref....
+    {                                             /////////////////       this would require BuffHandle to be a smart ref....     --> ///TICKET #249
       BuffHandle 
       step (Invocation& ivo)
         {

src/proc/engine/nodewiring.cpp

@@ -26,7 +26,7 @@
 #include "proc/engine/nodeoperation.hpp"
 #include "proc/engine/nodewiring-config.hpp"
 
-#include "lib/meta/typelist-util.hpp"
+#include "lib/meta/typelist-manip.hpp"
 
 
 namespace engine {

src/proc/engine/pluginadapter.hpp

@@ -24,7 +24,7 @@
 #ifndef ENGINE_PLUGINADAPTER_H
 #define ENGINE_PLUGINADAPTER_H
 
-#include "proc/engine/trafo.hpp"
+#include "proc/engine/procnode.hpp"
 
 
 
@@ -37,7 +37,7 @@ namespace engine
    * Effects processors are typically defined in a separate library and
    * will be loaded at runtime using Lumiera's plugin interface.
    */
-  class PluginAdapter : public Trafo
+  class PluginAdapter : public ProcNode
     {
       /////////////
     };

src/proc/engine/render-invocation.hpp

@@ -37,7 +37,7 @@
 //#include "proc/state.hpp"
 #include "proc/engine/procnode.hpp"
 #include "proc/engine/buffhandle.hpp"
-//#include "proc/engine/bufftable.hpp"
+//#include "proc/engine/bufftable-obsolete.hpp"