I asked this as the custom version isn't used for the Win32 port. The
makefile comments that the custom version is GTK+1.2, whereas the GTK+
version used for Win32 is 1.3. I should do a diff with the GTK+1.3 version
and see what's changed.
> -----Original Message-----
> From: owner-ethereal-dev@xxxxxxxx [mailto:owner-ethereal-dev@xxxxxxxx]On
> Behalf Of Gilbert Ramirez
> Sent: 03 May 2000 14:04
> To: Graham Bloice
> Cc: Guy Harris; ethereal-dev@xxxxxxxx
> Subject: Re: [ethereal-dev] UI Updates
>
>
> On Wed, May 03, 2000 at 09:53:43AM +0100, Graham Bloice wrote:
> > I suspect that the target over which the mouse button is
> released gives the
> > cue. If it's the same column label as the mouse down then it's
> a sort, else
> > it's a drag.
> >
> > If GTK+ adds the facility then I would expect it to emit a
> different signal
> > for the two operations.
> >
> > On that note why does ethereal have its own copy of clist ?
>
> It's our own modification of clist. We should have renamed it, to
> avoid symbol
> conflicts on Win32 and AIX. GTK+'s GtkCList adds rows via the
> g_list_append, which
> does *not* maintain a pointer to the end of the list. Each time
> an append is made,
> the entire GList of rows is searched for the tail node.
>
> For very large capture files, this exponential time became noticable.
>
> Our modified GtkCList keeps track of the tail GList node and
> appends directly to
> that.
>
> Guy has submitted the changes to the GTK+ folks, but they have
> not integrated
> his patches into GTK+. By a note I read on the mailing list, they
> seem to want
> to wait until GTK+-1.4 is out so they can use the GQueue
> container, which apparently
> is a GList that knows where its tail is.
>
> --gilbert
>
