Hi,
----- Original message -----
> On Tue, 2010-09-07 at 21:36 +0200, Jens Georg wrote:
>
> > currently GSSDP uses a home-brewn GSource implementation which is
> > neither portable nor 21st century ;)
> >
> > During the discussion with Zeeshan and Ross of the future of my
> > Windows port of GSSDP and why it cannot be merged currently, I
> > suggested that it would probably be more future-proven to port GSSDP
> > to GIO. That said, here's a pretty straight-forward port of my
> > windows port to GIO. I used my windows port as a starting point since
> > it already modifies some of the code in a GSocket friendly way (by
> > accident, actually). You can find the code here:
> >
> > http://gitorious.org/~phako/gupnp/jensg-gssdp/commits/gio
> >
> > It's still a bit rough around the edges (like might leak references)
> > but works well with gssdp-device-sniffer, gupnp-av-cp and Rygel. I
> > tagged the version which I consider feature-equivalent to current
> > master with gio-0.1.
> >
> > Nevertheless, there is room for more improvement (in no particular
> > order):
> > * Add some tests
> > * Make the SocketSource a GObject and move the event handling done in
> > the client into the SocketSource
> > * Make the multicast code nice enough to be included into glib.
> > * Create a portable physical network layer library which will include
> > the host-ip detection code of GSSDP and all of the network
> > management code currently in gUPnP. This will encapsulate all of
> > the network portability ugliness, keeping the other libs nice and
> > shiny
> > * Make it IPv6 aware.
> >
> > Please share your opinions on that.
As you probably know, I'm on vacation but I did briefly look into the patches last week and they looked good IIRC so if Sven and Ross are ok with it, I have no objections either.
--
Regards
Zeeshan Ali (Khattak)
Sent from Nokia N900