Some panel bug notes (Fwd: Re: panel bugs.)

[Telsa Gwynne <hobbit aloss ukuu org uk>]

This is from email between me and George Lebl. Some of it might be
useful for people.

The references to things like setup_menu_item and so on are to
particular functions found in the stack traces, where all the 
stack traces were line-for-line duplications. 

They're all from panel bugs, although many of them were found in
gnome-applets or general. 

> Date: Tue, 8 May 2001 02:24:49 -0700
> From: George <jirka 5z com>
> To: Telsa Gwynne <hobbit aloss ukuu org uk>
> Subject: Re: panel bugs.

> On Mon, May 07, 2001 at 10:55:26PM +0100, Telsa Gwynne wrote:
> > another menu_tear_off() crash. And here's another one with the 
> > launchers".
> > 
> > The most fun pairing: 
> > 53626: panel doesn't save launchers pulled from menus, but saves others
> > 53759: panel saves launchers pulled from menus, but doesn't save others
> 
> OK, figured this one out! Yay!  Actually if you study the bugreports
> carefully it seems that they were all really one thing and that was
> "launchers added from the Add to panel/Launcher from menu submenu
> don't persist".  The problem is people manage to misdescribe the
> problem in horrible ways :)
> 
> The new scheme in the panel is hoarding the desktop files for launchers into
> my own dir.  And I was forgetting to hoard those.
> 
> > setup_menu_item: all old (Added to 53833)
> 
> Closed as INCOMPLETE now :)
> 
> > init_user_panel: ooh, lots! (Added them all to 54068)
> 
> closed as INCOMPLETE now.  This is some bug in 1.0.55 that never shows 
> up in any other version.  But thanks to bugbuddy of 1.0.55 not requireing 
> people to even type in garbage, it results in thousands of identical 
> bugreports with just the stacktrace.
> 
> > menu_tearoff: even more! (Added to 54222)
> 
> This one is an interesting one.  Still trying to figure out what could be
> happening here.  I can't figure it out.  The calltree doesn't make sense
> since it's probably too optimized to be well debugged and doesn't have 
> enough symbols.  But the printf at the end gives a little bit of a clue 
> so this may be possible to kill perhaps.
> 
> > There's an "interesting" (allegedly) alpha crash at 54221.
> 
> Well, the stacktrace has NO interesting info in it.  I'll just close this 
> bug as INCOMPLETE at some point.

> > There's another two crashes blamed on the panel which are the
> > fault of mailcheck applet. I'm sorry. I should have sent you
> > the applet reports months ago. Mailcheck is evil.
> 
> Mailcheck crashes should be fixed.  They are usually the IMAP code which 
> was pretty buggy in 1.2.  I should look through those and close the relevant
> ones.
> 
> Actually if you think mailcheck was evil before, it's even eviler now.  I
> have some very cool evil code to make it asynchronious.
> 
> George
> -- 
> George <jirka 5z com>
>    How shall we fuck off, o Lord?
>                        -- Monty Python