Re: gtkmm 3.0 and libsigc++

[Chris Vine <chris cvine freeserve co uk>]

On Tue, 27 Jul 2010 12:06:40 +0200
klaus triendl <klaus triendl eu> wrote:
> Krzysztof Kosiński schrieb:
> > Hello
> > 
> > gtkmm 3.0 is the only chance we have in near future to fix two
> > really bad issues in libsigc++:
> > 1. sigc::trackable is not threadsafe.
> 
> For what exactly do you need a threadsafe sigc::trackable?
> 
> You can take a look at sigx++ for threadsafe interthread
> communication, which also allows for connecting to/disconnecting from
> signals in a threadsafe manner.

The proposal is not (I think) about replacing Glib::Dispatcher or
making new provision for inter-thread communication via main loops.  The
problem with sigc::trackable is that because the slot container is not
thread safe, once one thread has created a slot representing a
non-static method of an object deriving sigc::trackable, no other
thread can create a slot for any other method of the object.

It would be nice if signals were themselves thread safe, but the fact
that sigc::trackable is not thread safe is a major constraint in the
use of libsigc++ in multi-threaded programs.

Having said that, the documentation on sigx++ is quite limited so I am
not sure to what extent it provides thread-safe intra-thread operation
(which is what we are talking about) rather than inter-thread operation
(which is what the sigx++ home page seems to major on) with reasonable
speed and efficiency.  Does sigx++ normally send everything through a
main loop?

Chris