[gnet-dev] RE: [gtkmm] Re: gnet-library - C++-bingings?

[Murray Cumming Comneon com]

> From: Tassos Bassoukos [mailto:abas aix meng auth gr]
> [snip part about my attempt at gnet C++ wrappers]
> 
> > Maybe you could send me them anyway? So I could learn by 
> example how to
> > do that stuff.
> > 
> > Thanks,		Jonas
> 
> Hi Jonas,
> Are you sure? I've just re-read them and they stink[0].

It doesn't look that bad to me:
http://www.gnetlibrary.org/docs/

> So far, in my attempts to wrap gnet i've found out these facts:
> 1. while h2defs[1] works on the gnet headers, the resulting defs file
>     needs extensive reorganization, as gnet is (mostly) not
>     GObject-based.

It's been a long time since I've seen h2defs fail on something. Could you
tell me more about the specific problems? h2defs can be fixed, or gnet can
be corrected.

It might not make sense for them to be GObject-based. For instance, maybe
inheritance is not necessary.

> 2. seral "interesting" complexities exist on the async 
> interfaces,

Examples please? URLs for reference documentation or web cvs would be best.

> which
>     would require understanding of SigC::Connection and friends..

A simple SignalProxy (like the ones in TreeView) is usually all that's
required to map from function pointers to SigC::Slots.

> 3. Oh, and it works very nicely with the event loop :-)

That's good, right?

Murray Cumming
murrayc usa net
www.murrayc.com