'make check' failure

[Michael Stahl]

On 23/08/12 10:19, Stephan Bergmann wrote:
> On 08/23/2012 09:22 AM, John Smith wrote:
>> On Thu, Aug 23, 2012 at 9:14 AM, Stephan Bergmann <sbergman at redhat.com> wrote:
>>>
>>> you should always run those tests with (bash etc.:) 'ulimit -c unlimited';
>>> the gbuild logic will automatically print backtraces in case of a crash then
>>> (if your system is set up to generate core files named "core" or
>>> "core.<pid>" into cwd).
>>>
>>> Stephan
>>>
>> I do have core set to unlimited, but I still got a message that a core
>> could not be generated, and all the output I got was what I posted
>> already.
> 
> Odd then.  Maybe your system's behavior doesn't match those "if your 
> system..." requirements from above?

reading the log you posted no you don't get a core file because
soffice.bin didn't actually crash.

the test framework that runs unoapi tests will happily continue running
more tests until the end if one simply fails to give the expected results.

i think i've seen a failure in some ScAccessibleFoo as well a week or
two ago on Linux, but couldn't reproduce it.