Commit graph

3479 commits

Author SHA1 Message Date
e76b85fcfa fix test broken by #410 2012-10-10 05:20:19 +02:00
0e3821abeb refatcor function meta programming headers
the template _Fun started as an internal helper
for function-closure, but seems to be of
general use. Thus move it into meta/function.hpp
(function-closure.hpp is heavyweight)
2012-10-10 05:20:19 +02:00
e6a105fbc1 iterator exploring monad finished, passes unit test 2012-10-10 05:20:19 +02:00
3d0d599158 get the depth-first exploration test to work
...using the IterQueue for intermediary results
2012-10-10 05:20:18 +02:00
2c33000346 complement IterStack by a similar wrapper for queue-like access 2012-10-10 05:20:18 +02:00
8ced6758fb initial draft for recursive evaluating iterator monad
tricky design problem, because nothing is known about
the source and result sequences to be built.
2012-10-10 05:20:18 +02:00
59c5627bbe change Implementation class for IterStack
now using a std::deque directly; this allows
for a later extension for queue-like behaviour
2012-10-10 05:20:18 +02:00
41180eb99c rework to allow for recursive evaluation
this enables expansion of a (functional) data structure
until exhaustion -- which is what we need to
build job functors by traversing and expanding
an arbitrarily nested job definition structure
2012-10-10 05:20:17 +02:00
75df583607 raw version of ChainedIter passes unit test 2012-10-10 05:20:17 +02:00
697924925c pragmatic fix for the long-standing problem of detecting increment operator
our duck-detector had a shortcoming here and failed
to detect iterators, in case the increment operator
was inherited from a mix-in
2012-10-10 05:20:17 +02:00
8db5413199 implemented chained-iterators
...using the IterExplorer building blocks
2012-10-10 05:20:17 +02:00
d14b37a71d utility class factored out and covered by test
...a stack which can be Lumiera-iterated
2012-10-10 05:20:16 +02:00
da4a343e9e refactor IterExplorer to allow for more flexible strategy definition 2012-10-10 05:20:16 +02:00
dc3ebd4a8f first working concept for an "iterator monad"
the intention is to use this to simplify
generating render jobs based on the elaborated
dependency network of the render nodes. The key
challenge is to overcome the necessity to
store partially done evaluations as
continuation
2012-10-10 05:20:16 +02:00
3a7db1603e add test for the base case 2012-10-10 05:20:16 +02:00
eedcd69941 draft the IterExplorer design
the tricky part seems to be how to combine the
source iterators into a new monad instance, while
keeping this "Combinator" Strategy configurable

...just passes the compiler, while still lacking
even the generic implementation of joining
together the source iterators
2012-10-10 05:20:16 +02:00
45f4c96c6f change LinkedElements to us a more space efficient iterator
Actually we don't need any backreference to the
container for iterating a singly linked list
2012-10-10 05:20:15 +02:00
378ebe21f0 Fix naming of Iteration control API functions (closes #410)
comes in handy now, since IterStateWrapper uses a similar API
2012-10-10 05:20:15 +02:00
83a0f4b41f Implementation (I) : IterStateWrapper as foundation (passing test) 2012-10-10 05:20:15 +02:00
ab2a6b2fce WIP brainstorming about a monadic iterator
The idea is to avoid building a data structure
for intermediary results, while still being able
to process a variably sized and arbitrary shaped
set of source data
2012-10-10 05:20:15 +02:00
fa15a38016 draft JobTicket implementation datastructure
...but need an extension on our iterator helpers
to build a suitable functional datastructure
for generating Jobs.
2012-10-10 05:20:15 +02:00
3ef1fb7697 linked list helper template finished and passes test 2012-10-10 05:20:14 +02:00
c33fcf9797 WIP draft a linked list helper template 2012-10-10 05:20:14 +02:00
ddff8b654b WIP investigating the relation of Jobs, JobTicket and Closure in detail 2012-10-10 05:20:14 +02:00
4ceec90b9e design decision how to invoke a job
without any trickery, we'll always get two indirections.
Thus, the decision is to turn the 2nd indirection
into a VTable call; this way, basically the JobClosure
also acts as job functor itself.
2012-10-10 05:20:13 +02:00
0320bc4b2c considering the relation of Job and JobClosure 2012-10-10 05:20:13 +02:00
bb43c03ef9 stub all the job generating functions required for the dispatcher interface 2012-10-10 05:20:13 +02:00
0ab773ab7c helper to pull all elements from an iterator, yielding the last one 2012-10-10 05:20:13 +02:00
5222384f8b try to circumvent ambiguity in overload resolution
we want to support various integral and rational factors
for linear chaining of time offsets
2012-10-10 05:20:13 +02:00
79bd8b71e3 better treat the building of a continuation job separately 2012-10-10 05:20:13 +02:00
08d266819d re-read my own code and pick up the design work
..next question is: how to shape the dispatcher interface,
in order to support ongoing chunk wise planning
of new jobs, including a continuation
2012-10-10 05:20:13 +02:00
080d6913c6 fix a type converison problem
...better state explicitly we're requiring
a rational factor
2012-10-10 05:20:12 +02:00
7941865d5d implement anchor against current system time
using CLOCK_REALTIME for now
2012-10-10 05:20:12 +02:00
9aec2a9806 allow for fractional scaling of time durations
implemented as extension to the linear combinations.
I decided to use the same "always floor" rule
as employed for time quantisation. Moreover,
we don't support floating point, only rationals
2012-10-10 05:20:12 +02:00
22322dfec4 refactor the division/quantisation helpers
...no need to keep them in util.hpp, as they
are used rather occasionally, while util.hpp
is used pervasively.
2012-10-10 05:20:12 +02:00
882bcf07ae define stubs for accessing the wall clock time (->Backend) 2012-10-10 05:20:12 +02:00
aa01813f52 TimeAnchor: draft calculating the start delay for rendering 2012-10-10 05:20:12 +02:00
ee1450a81a rectify frame dispatch invocation 2012-10-10 05:20:12 +02:00
393a447861 test setup: simulated build proecss results, mock model port 2012-10-10 05:20:12 +02:00
f8f011bb44 rework Job representation
make class Job a real subclass of the
job definition struct and turn the
JobClosure into a trampoline
2012-10-10 05:20:12 +02:00
0b25c2e08d Fix: missing sanity check in ScopedCollection
funny enough this possible memory corruption
didn't happen in the unit test, because my
compiler optimised the additional int field
of class SubDummy, making it the same size
of the baseclass. Now matters should be safe.
2012-10-10 05:20:12 +02:00
1e54b5d3e6 stubbing some job functions 2012-10-10 05:20:11 +02:00
365f54d59e Proc-Layer TiddlyWiki entrance pages: some minor updates 2012-10-10 05:20:11 +02:00
db68577b4a clarify relation of Job, JobTicket and channel number 2012-10-10 05:20:11 +02:00
875342fa40 initial draft for the job representation
this draft is based on
 - Cehteh's draft for the scheduler
 - my plannings about segmentation and JobTicket

it defines "Job" as a closure which can be invoked
from plain-C, using the information in the
job descriptor datastructure
2012-10-10 05:20:11 +02:00
e9dbb3bdb1 stubs for some important components of play/engine (JobTicket...)
also touches the question how to represent the job
descriptor datastructure. @Cehteh: I've just pasted
in your preliminary data struct definitinons
from the relevant mailing list discussions.
2012-10-10 05:20:11 +02:00
23ac29028c design draft: job tickets and planning... 2012-10-10 05:20:10 +02:00
568fadd526 draft some steps of the dispatch operation 2012-10-10 05:20:10 +02:00
6772e94994 implement simple constant frame timings descriptor 2012-10-10 05:20:10 +02:00
157e3b6867 test-driven brainstorming: define default timings... 2012-10-10 05:20:10 +02:00