[webkit-dev] run-webkit-tests failure

[Eric Seidel]

open /Developer/Applications/Utilities/CrashReporterPrefs.app

will allow you to at least change your Crash Reporter preferences to  
show you when applications crash ("Developer").

Enabling coredumps is a little more tricky under Mac OS X.  I believe  
under tiger, you have to do the following:

echo "limit core unlimited" >> /etc/launchd.conf

but there may be more steps.

-eric


On Sep 11, 2005, at 2:50 PM, Kurt Kohler wrote:

> Sorry that I wasn't very clear.
>
> Actually I was wrong, only one of the leaks I saw looked like it was
> from DumpRenderTree. What made me think that was the fact that the  
> leak
> was the string "DumpRenderTree". Is there some other reason that would
> show up? I admit I don't understand all this stuff very well yet.
>
> I'll watch to see if there's a way to get a core dump when
> DumpRenderTree crashes. It's moderately strange that the crash appears
> to be random. In my experience that often means some variable isn't
> getting initialized.
>
> Kurt
>
> Darin Adler wrote:
>
>> On Sep 7, 2005, at 11:59 AM, Kurt Kohler wrote:
>>
>>> I've noticed that occasonally the run-webkit-tests script fails   
>>> with
>>> messages like
>>>
>>> css1/basic/class_as_selector.html -> failed
>>> .Broken pipe
>>>
>>> When I run it again it works fine. Is there any significance to  
>>> this?
>>
>>
>> This means DumpRenderTree crashed. I wonder if there's a way to see a
>> backtrace of the crash. If not, we should fix the script.
>>
>>> By the way, I see that the only leaks left appear to be from
>>> DumpRenderTree. Does that mean all the leaks in webkit itself are
>>> fixed? If so, congratulations!
>>
>>
>> I'm not sure what you mean. Leaks from DumpRenderTree itself? That
>> sounds surprising to me. Are there any Bugzilla bugs about these
>> leaks you're talking about?
>>
>> We have definitely fixed a lot of leaks, and the vast majority of the
>> ones that show up when running the layout tests, but I do see some
>> leaks from time to time.
>>
>>     -- Darin
>
> _______________________________________________
> webkit-dev mailing list
> webkit-dev at opendarwin.org
> http://www.opendarwin.org/mailman/listinfo/webkit-dev