Re: [gene-pool] Re: soup patch, 29731

On Tue, 2002-10-29 at 01:50, Dan Winship wrote:
> On Mon, 2002-10-28 at 01:09, Not Zed wrote: 
> > Patch addresses 29731, I'm not sure all cases are completely handled,
> > for example a successful immediate connect, should that return NULL or
> > not?
> 
> It should return NULL, since you can't cancel the attempt at that point.

Ok.

> >   But otherwise it fixes the crash in 29371, tho maybe it could use
> > some neatening (e.g. unused enum's).
> 
> Alternate patch attached; it just uses state->tcp_id and
> state->inetaddr_id to distinguish the immediate callback case from the
> async case, which makes it clearer I think.

Larry, can you check this?  He's been running with the other patch and
had a lot more success with evolution than he had been.  I'm a bit too
rooted to check it, tho it looks ok at first glance.

> > There is also a small patch to fix another problem with cancelled
> > connections never being discounted, which means after 10 cancellations,
> > connections are never re-initiated.  Which basically stops image loading
> > working at all.
> 
> That part looks good.

Its not quite complete however.  There are a couple of other cases where
the count isn't decremented.  I'm not sure where they are though :-/ 
But they dont happen anywhere near as often, so its worth it as it
stands.

> > And finally a patch (request of Larry) to turn of the PTRACE debugging,
> > as it just hinders doing any debugging with it turned on.
> 
> Hm... Alex added the ptrace code because otherwise setting breakpoints
> on certain functions (like soup_gethostbyname or g_strdup) would make
> hostname lookups fail reliably. It must not work right with multiple
> threads or something.

Larry can comment on this i'm sure ... when he eventually wakes up.



_______________________________________________
Soup-list maillist  -  [email protected]
http://lists.ximian.com/mailman/listinfo/soup-list