bug#15088: 24.3; Hard crash on OSX when editing header files

[Simon Stapleton]

Yeah, my guess is package version.  Both binaries (old and newly compiled), 
plus several older compiles, suffer from the crash issue.  With newly pulled 
packages, no crash.

Thanks for the help anyway.

Simon

On 16 Aug 2013, at 07:07, Jan Djärv <jan.h.d@swipnet.se> wrote:

> Hello. 
> 
> I guess we can keep this bug report in case it happens again.  I don't think 
> looking at your setup will help, these things are usually compiler and/or 
> package version dependent. 
> 
>      Jan D.
> 
> 14 aug 2013 kl. 20:23 skrev Simon Stapleton <simon.stapleton@gmail.com>:
> 
>> Hi Jan
>> 
>> Nope, nothing at all.  So I started trying to track down what was causing 
>> the issue.  Starting from a virgin .emacs.d/ and .emacs, started re-adding 
>> packages to find out which one actually breaks things.
>> 
>> And, wouldn't you know it, I'm pretty much back to where I was before, and 
>> it no longer happens.  The only thing I haven't reinstalled thus far is 
>> quack, which appears to no longer be in elpa/melpa, and shouldn't be 
>> touching C-mode files anyway.
>> 
>> This confuses me no end, as I'd started from a virgin .emacs / emacs.d when 
>> I moved to 24.3, and re-zapped them a few times trying to get rid of this 
>> very problem.
>> 
>> I can drop a zip/tarball of my old .emacs stuff if it would help anyone 
>> figuring out what's going on.
>> 
>> 
>> 
>> 
>> 
>> On 13 Aug 2013, at 22:29, Jan Djärv <jan.h.d@swipnet.se> wrote:
>> 
>>> Hello.
>>> 
>>> 13 aug 2013 kl. 20:58 skrev Simon Stapleton <simon.stapleton@gmail.com>:
>>> 
>>>> Okay, quick rebuild and I can now get a full backtrace.  xbacktrace still 
>>>> fails, but it's failing because (I assume) it's had its stack smashed
>>>> 
>>>>> No symbol "ARRAY_MARK_FLAG" in current context. 
>>>> 
>>>> ARRAY_MARK_FLAG looks a lot like a garbage collection bug being tickled, 
>>>> double release going on somewhere maybe?
>>>> 
>>>> The source of the issue is almost certainly to do with pabbrev - 
>>>> commenting it out in my .emacs makes everything work.
>>> 
>>> A lisp package should not be able to crash Emacs.  You can try to set these 
>>> in your environment before starting Emacs from the command line:
>>> 
>>> export NSZombieEnabled=YES
>>> export MallocGuardEdges=1
>>> export MallocScribble=1
>>> export MallocErrorAbort=1
>>> export MallocCorruptionAbort=1
>>> export MallocCheckHeapStart=1000
>>> 
>>> Emacs will run very slowly, but hopefully it will print something useful if 
>>> it is a memory issue.
>>> 
>>>   Jan D.
>>> 
>>>