Re: [gnet] Re: GNet library (2.0.5)

[David Helder <dhelder gizmolabs org>]

The last time we discussed this I was anti-GError because I wanted to 
maintain compatibility with GLib 1.  But I think we can safely drop 
this requirement.

David


On Nov 24, 2004, at 6:13 AM, Juergen Schoenwaelder wrote:

> On Wed, Nov 24, 2004 at 10:05:53AM +0000, Tim M?ller wrote:
>
> > > For callbacks lets simply pass an GError * as well.
> >
> > One problem with GError * is that it only exists in GLib-2.0, but not 
> > in
> > GLib-1.2.
> >
> > Personally, I think it's time for GNet to drop support for GLib-1.2
> > altogether, and to start making more extensive use of all the GLib-2.0
> > goodies. At some point we will need to start supporting non-ASCII 
> > domain
> > names (RFC3490), and the GLib-2.x unicode stuff will come in handy 
> > for that.
>
> I would love to see usage of GError and I personally do not have a
> problem to require glib 2.0. I have an SNMP engine written on top of
> glib/gnet (dervived from gxsnmp code which again is derived to some
> extend from btng code) and I stared to add GError arguments everywhere
> so that low-level decoding errors can actually appropriately be 
> reported
> and handled. This is really worth the effort and I would love if the
> gnet library functions that deal with socket would also give me usable
> error information which I can forward using my API functions.
>
> [Right now, I have to fake GError messages for things such as
>  gnet_tcp_socket_new() or gnet_udp_socket_send() and this is ugly
>  at best.]

--
David Helder - dhelder gizmolabs org - http://www.gizmolabs.org/~dhelder