|
|
Boost Testing : |
Subject: Re: [Boost-testing] Bad commit?
From: Belcourt, K. Noel (kbelco_at_[hidden])
Date: 2010-12-05 15:50:44
On Dec 5, 2010, at 1:41 PM, Jeremiah Willcock wrote:
> On Sun, 5 Dec 2010, Belcourt, K. Noel wrote:
>
>>
>> On Dec 5, 2010, at 1:34 PM, Jeremiah Willcock wrote:
>>
>>> On Sun, 5 Dec 2010, Belcourt, K. Noel wrote:
>>>
>>>>
>>>> On Dec 5, 2010, at 1:27 PM, Jeremiah Willcock wrote:
>>>>
>>>>> On Sun, 5 Dec 2010, Jeremiah Willcock wrote:
>>>>>
>>>>>> On Sun, 5 Dec 2010, Belcourt, K. Noel wrote:
>>>>>>> It appears all trunk testing is broken due to a change affecting
>>>>>>> process_jam_log.cpp. The bad commit likely occurred sometime
>>>>>>> Friday as
>>>>>>> Thursday night's tests were okay. Here's the error message.
>>>>>>
>>>>>> I think this might be from the change to Boost.Filesystem v3 as
>>>>>> the
>>>>>> default
>>>>>> version. I had some BGL failures that had similar messages and
>>>>>> were
>>>>>> fixed
>>>>>> by
>>>>>> updating the Filesystem code.
>>>>>
>>>>> I attached a patch that fixes the problem on my system (with
>>>>> GCC). I can
>>>>> commit it directly if you want.
>>>>
>>>> Yes, please do Jeremiah. That's a big help.
>>>
>>> Do you have a way to test it? I think I may not have things set up
>>> properly to run the test suite. Do you want me to commit it then
>>> you try
>>> it out?
>>
>> Well my thinking is that all trunk testing is broken if you
>> actually compile
>> process_jam_log from source, so committing it to trunk won't make
>> things any
>> worse. And yes, if you commit it, I promise to login and test it
>> out to make
>> sure it works.
>
> OK -- I did that in r67039.
Works like a champ.
[kbelco_at_wsblade001 results]$ ls -l /scratch/boost/tools_regression/
build/bin/gcc-4.1.2/release/process_jam_log
-rwxrwxr-x 1 kbelco kbelco 163419 Dec 5 13:48 /scratch/boost/
tools_regression/build/bin/gcc-4.1.2/release/process_jam_log
The Sandia-intel-11.0-64 tester will be the first Sandia tester to
cycle.
You're awesome Jeremiah, thanks!
-- Noel