|
|
Boost Testing : |
Subject: Re: [Boost-testing] Result statuses for run failures of MPI tests
From: Jeremiah Willcock (jewillco_at_[hidden])
Date: 2010-07-04 16:39:42
On Mon, 31 May 2010, Belcourt, Kenneth wrote:
> Hi Stevem
>
> On May 31, 2010, at 9:25 AM, Steven Watanabe wrote:
>
>> Belcourt, Kenneth wrote:
>>>
>>> On May 30, 2010, at 5:56 PM, Jeremiah Willcock wrote:
>>>
>>>> On some platforms and compilers, several of the Boost.Graph.Parallel
>>>> (libs/graph_parallel/test) tests compile and link correctly but
>>>> produce a
>>>> failure when run. There is no information in these cases, either in the
>>>> form of program output or an error code. One example of this is
>>>> <URL:http://tinyurl.com/39noasn>. Is there more information
>>>> available to
>>>> the test scripts but not reported? Is there a way to at least get an
>>>> exit
>>>> status or something from the tests? There are similar failures on
>>>> Windows
>>>> as well, so an error/return code there would be useful as well.
>>>> Thank you
>>>> for any help you can give in tracking these down.
>>>
>>> That is strange. This test runs to completion just fine.
>>>
>>> <snip>
>>> EXIT STATUS: 0
>>>
>>> and the return code is zero. Not sure why the test is being reported
>>> as a fail. Perhaps this is a process_jam_log failure as it's clearly
>>> not a test failure.
>>
>> Hmmm. process_jam_log seems to produce
>> <run result="*succeed*" timestamp="*2010-05-31 07:16:08 UTC*" />
>>
>> Could it be that the fact that no output is shown is confusing
>> the report XSL stylesheets?
>
> Seems likely since, as you point out, PJL is writing success into the log.
> Unfortunately I'm no XSL expert, any ideas how to debug or patch this?
This issue is still appearing, in particular on Visual C++. Would it be
possible for someone to look at the logs for that platform and see if the
tests are actually failing? Also, are the raw test logs available
anywhere? I might be able to look through them myself to see if there is
a real bug in BGL or not.
-- Jeremiah Willcock