Recent enabling of xassert

Recent enabling of xassert

[Kim F. Storm]

Miles recently enabled xassert unconditionally to catch undiscovered
bugs in redisplay and other places...

We have now seen several crashes caused by this change -- but none of
them has been fully diagnosed yet, as this is very hard without having
easy ways to reproduce the crashes.

Debugging those crashes requires time which I don't have -- so unless
someone else steps in to help, I suggest we disable xassert again.

Since CVS emacs has run pretty flawlessly for years without those
xasserts (except for one recent glitch), it would at least be better
than the current situation!

That said, it would be fine if some users continue to run with xassert
enabled.

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Recent enabling of xassert

[Jason Rumney]

Kim F. Storm wrote:

>Miles recently enabled xassert unconditionally to catch undiscovered
>bugs in redisplay and other places...
>
>We have now seen several crashes caused by this change -- but none of
>them has been fully diagnosed yet, as this is very hard without having
>easy ways to reproduce the crashes.
>
>Debugging those crashes requires time which I don't have -- so unless
>someone else steps in to help, I suggest we disable xassert again.
>  
>
I disagree. xassert should only be disabled for releases. We want to 
find the bugs now, even if it takes time to fix them. CVS should not be 
considered a stable release.

Re: Recent enabling of xassert

[Kim F. Storm]

Jason Rumney <jasonr@gnu.org> writes:

> Kim F. Storm wrote:
>
>>Miles recently enabled xassert unconditionally to catch undiscovered
>>bugs in redisplay and other places...
>>
>>We have now seen several crashes caused by this change -- but none of
>>them has been fully diagnosed yet, as this is very hard without having
>>easy ways to reproduce the crashes.
>>
>>Debugging those crashes requires time which I don't have -- so unless
>>someone else steps in to help, I suggest we disable xassert again.
>>  
>>
> I disagree. xassert should only be disabled for releases. We want to
> find the bugs now, even if it takes time to fix them. CVS should not
> be considered a stable release.

I agree, but...

Proving the presense of bugs is no aim in itself (I'm sure there are
many bugs, but if they are "mostly harmless" it is not critical to
find them).

Why detect bugs if we don't have time to fix them (or diagnose
them as false xasserts)?

Actually, my suggestion was mostly rethorical -- asking/provoking
people to help debugging these (potentiall very hard to diagnose) bugs
as they are reported.

So who can and will help?

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: Recent enabling of xassert

[Eli Zaretskii]

> From: storm@cua.dk (Kim F. Storm)
> Date: Thu, 10 Feb 2005 12:15:44 +0100
> Cc: emacs-devel@gnu.org
> 
> > I disagree. xassert should only be disabled for releases. We want to
> > find the bugs now, even if it takes time to fix them. CVS should not
> > be considered a stable release.
> 
> I agree, but...
> 
> Proving the presense of bugs is no aim in itself (I'm sure there are
> many bugs, but if they are "mostly harmless" it is not critical to
> find them).

If some places where xassert is used are revealing harmless
situations, let's remove those xassert's.