When some dependency or singleton violates Lumiera's policy regarding destructors and shutdown,
we are unable to detect this violation reliably and produce a Fatal Error message.
This is due to lib::Depend's de-initialisating being itself tied to template generated
static variables, which unfortunately have a visibility scope beyond the translation unit
responsible for construction and clean-up.
- state-of-the-art implementation of access with Double Checked Locking + Atomics
- improved design for configuration of dependencies. Now at the provider, not the consumer
- support for exposing services with a lifecycle through the lib::Depend<SRV> front-end
...which declare DependencyFactory as friend.
Yes, we want to encourrage that usage pattern.
Problem is, std::is_constructible<X> gives a misleading result in that case.
We need to do the instantiation check within the scope of DependencyFactory
ideally we want
- just a plain unique_ptr
- but with custom deleter delegating to lib::Depend
- Depend can be made fried to support private ctor/dtor
- reset the instance-ptr on deletion
- always kill any instance
all these tests are ported by drop-in replacement
and should work afterwards exactly as before (and they do indeed)
A minor twist was spotted though (nice to have more unit tests indeed!):
Sometimes we want to pass a custom constructor *not* as modern-style lambda,
but rather as direct function reference, function pointer or even member
function pointer. However, we can not store those types into the closure
for later lazy invocation. This is basically the same twist I run into
yesterday, when modernising the thread-wrapper. And the solution is
similar. Our traits class _Fun<FUN> has a new typedef Functor
with a suitable functor type to be instantiated and copied. In case of
the Lambda this is the (anonymous) lamda class itself, but in case of
a function reference or pointer it is a std::function.
...which showed up under high system load.
The initialisation of the member variables for the check sum
could be delayed while the corresponding thread was already running
- polish the text in the TiddlyWiki
- integrate some new pages in the published documentation
Still mostly placeholder text with some indications
- fill in the relevant sections in the overview document
- adjust, expand and update the Doxygen comments
TODO: could convert the TiddlyWiki page to Asciidoc and
publish it mostly as-is. Especially the nice benchmarks
from yesterday :-D
This solution is considered correct by the experts.
Regarding the dependency-configuration part, we do not care too much about performance
and use the somewhat slower default memory ordering constraint
This is essentially the solution we used since start of the Lumiera project.
This solution is not entirely correct in theory, because the assignment to the
instance pointer can be visible prior to releasing the Mutex -- so another thread
might see a partially initialised object
_not_ using the dependency factory, rather direct access
- to a shared object in the enclosing stack frame
- to a heap allocated existing object accessed through uniqe_ptr
The Lumiera thread-wrapper accepts the operation to be performed
within the new thread as a function object, function reference or lambda.
Some of these types can be directly instantiated in the threadMain
function, and thus possibly inlined altogether. This is especially
relevant for Lambdas. OTOH, we can not instantiate function references
or bound member functions; in those cases we fall back to using a
std::function object, possibly incurring heap allocations.
the operation to run within the thread can be passed through as lambda;
and with the help of C++11 perfect forwarding we can get rid of the
temporary ThreadStartContext entirely, which removes a lot of
pass-through argument copying
...written as byproduct from the reimplementation draft.
NOTE there is a quite similar test from 2013, DependencyFactory_test
For now I prefer to retain both, since the old one should just continue
to work with minor API adjustments (and thus prove this rewrite is a
drop-in replacement).
On the long run those two tests could be merged eventually...
This is a complete makeover of our lib::Depend and lib::DependencyFactory templates.
While retaining the basic idea, the configuration has been completely rewritten
to favour configuration at the point where a service is provided rather,
than at the point where a dependency is used.
Note: we use differently named headers, so the entire Lumiera
code base still uses the old implementation. Next step will be
to switch the tests (which should be drop-in)
explicit friendship seems adequate here
DependInject<SRV> becomes more or less a hidden part of Depend<SRV>,
but I prefer to bundle all those quite technical details in a separate
header, and close to the usage
This is a tricky problem an an immediate consequence of the dynamic configuration
favoured by this design. We avoid a centralised configuration and thus there
are no automatic rules to enforce consistency. It would thus be possible
to start using a dependency in singleton style, but to switch to service
style later, after the fact.
An attempt was made to prevent such a mismatch by static initialisiation;
basically the presence of any Depend<SRV>::ServiceInstance<X> would disable
any usage of Depend<SRV> in singleton style. However, such a mechanism
was found to be fragile at best. It seems more apropriate just to fail
when establishing a ServiceInstance on a dependency already actively in
use (and to lock usage after destroying the ServiceInstance).
This issue is considered rather an architectural one, which can not be
solved by any mechanism at implementation level ever
up to now we used placement into a static buffer.
While this approach is somewhat cool, I can't see much practical benefit anymore,
given that we use an elaborate framework which rules out the use of Meyers Singleton.
And given that with C++11 we're able just to use std::unique_ptr to do all work.
Moreover, the intended configurability will become much simpler by relying
on a _closure_ to produce a heap-allocated instance for all cases likewise.
The only possible problem I can see is that critical infrastructure might
rely on failsafe creation of some singleton. Up to now this scenario
remains theoretical however
Meyers Singleton is elegant and fast and considered the default solution
However...
- we want an "instance" pointer that can be rebound and reset,
and thus we are forced to use an explicit Mutex and an atomic variable.
And the situation is such that the optimiser can not detect/verify this usage
and thus generates a spurious additional lock for Meyers Singleton
- we want the option to destroy our singletons explicitly
- we need to create an abstracted closure for the ctor invocation
- we need a compiletime-branch to exclude code generation for invoking
the ctor of an abstract baseclass or interface
All those points would be somehow manageable, but would counterfeit the
simplicity of Meyers Singleton
Problems:
- using Meyers Singleton plus a ClassLock;
This is wasteful, since the compiler will emit additional synchronisation
and will likely not be able to detect the presence of our explicit locking guard
- what happens if the Meyers Singleton can not even be instantiated, e.g. for
an abstract baseclass? We are required to install an explicit subclass configuration
in that case, but the compiler is not able to see this will happen, when just
compiling the lib::Depend
Most dependencies within Lumiera are singletons and this approach remains adequate.
Singletons are not "EVIL" per se. But in some cases, there is an explicit
lifecycle, managed by some subsystem. E.g. some GUI services are only available
while the GTK event loop is running.
This special case can be integrated transparently into our lib::Depend<TY> front-end,
which defaults to creating a singleton otherwise.
we'll use a typedef to represent the default case
and provide the level within the UI-Tree as template parameter for the generic case
This avoids wrapping each definition into a builder function, which will be
the same function for 99% of the cases, and it looks rather compact and natural
for the default case, while still retaining genericity.
Another alternative would have been to inject the Tree-level at the invocation;
but doing so feels more like magic for me.
