On Thu, Dec 12, 2002 at 07:05:03PM +0100, J.Smith wrote:
> >
> > I didn't see that warning from 0.9.8 on Windows 2000.
> >
> That seems odd. I get it almost immediately on Win98SE (on two separate
> systems) when I have loaded the tracefile I had attached earlier.
Well, perhaps there's some difference between Windows OT (95/98/Me) and
Windows NT (NT 4.0/2K/XP/.NET Server) that causes it to happen on OT but
not NT.
> I don't
> have a Win2k/XP installation available to me to try it out on. Please note
> that I do not get it on all RPC session traces, but the session I supplied
> earlier did reproducibly trigger it for me.
I tried it with that trace, doing exactly the stuff you mention.
> > GTK+ is reporting a problem; whether the problem it's reporting is the
> > result of a GTK+ bug or an Ethereal bug is another matter.
> >
> Ok, so how do I go about figuring out which it is ?
I don't know.
> 2.) Since im not a developer but just a humble sysadmin ive never been able
> to get a debugger to produce anything more serious than a simple backtrace,
> so in order to produce anything that's more complex I would require some
> serious help from a developer if anyone would like me to set breakpoints and
> 'step-through-the-code'.
Unfortunately, I don't have time to supply that help, so, unless
somebody else does, you might just have to live with that message
popping up.
> Are there any alternatives that might help out here? Is there a way to start
> ethereal producing more verbose debugging output ?
None that I know of - there might be some way to provoke GTK+ to produce
more information, but I don't have time to look into that.
> Is there a way to force
> it to produce a crash when the assertion fails and submit that output to the
> list ?
I don't know.
> Is there another free debugger available for Win32 (that anyone on
> this list knows how to use) besides gdb ?
I don't know of one. On those occasions when I debug the Windows
version of Ethereal, I use the debugger that came with my copy of
MSVC++.
> > "gtk_widget_map()" isn't called by any Ethereal code, except by our
> > internal version of the CList widget (which is there because older
> >
> Just a long shot here - but would it theoretically be possible for
> differences that might possibly exist between these versions of this "CList
> widget" be the cause of this issue ?
It's probably not the result of our changes to that widge - the current
version of the widget is, I think, identical to the current GTK+ 1.2.10
version of the widget.
> > I think we're also distributing an older
> > version of the 1.3.x build.
> >
>
> Does this imply that the precompiled Ethereal is compiled/linked against a v
> ersion of GTK that's older than the version that's supplied on the GTK/win32
> website, and that I am to expect problems when I use that newer version ?
It might be the case.