hg-1.4.2 crash

[Mads Kiilerich]

Neal Becker wrote, On 01/17/2010 02:54 PM:
> Mads Kiilerich wrote:
>
>    
>> Neal Becker wrote, On 01/17/2010 02:31 AM:
>>      
>>> A user got the following backtrace.  Any clues what could be the cause?
>>>
>>> https://bugzilla.redhat.com/show_bug.cgi?id=556124
>>>
>>>        
>> https://bugzilla.redhat.com/show_bug.cgi?id=556249 comes with a
>> (unnecessarily complex) way to reproduce.
>>
>> The problem is what happens if Mercurials stderr has been closed and
>> Mercurial gets killed and wants to issue the "killed!" warning.
>>
>>   From one point of view: Nobody can see the crash, so why not just let
>> it crash? What should it do instead of crashing?
>>
>>   From another point of view: There was a crash and the system crash
>> catcher notices it and reports it. To avoid false reports we should
>> avoid crashing when known problems like this is met.
>>
>> /Mads
>>      
> I understand stderr was closed, but what killed mercurial?

I suggest asking the user what killed mercurial.

One of SIGBREAK, SIGHUP and SIGTERM was caught. That signal must have 
been sent by something else - perhaps the user did that intentionally, 
or perhaps he logged off while hg was cloning.

> Was this just a
> normal condition, or was there actually some error?
>    

Hmm ... I would call it a normal error condition ;-)

Mercurial could however handle it more elegant.

/Mads