Regression tests machines
I am managing some spare machine with reasonable performances and compiler selection that are basically underused at the moment (including various i64, AMD and PowerPC server blade) with both gcc and icc. I guessed they could be used to run the boost regression test on a dialy or weekly basis. I found out about the : http://www.boost.org/development/running_regression_tests.html page but I wanted to know if there is any speficic need in term of paltofrm/compiler coverage or if I can just run whatever and push the regression result up ?
On Jun 30, 2010, at 2:07 AM, Joel Falcou wrote:
I am managing some spare machine with reasonable performances and compiler selection that are basically underused at the moment (including various i64, AMD and PowerPC server blade) with both gcc and icc.
I guessed they could be used to run the boost regression test on a dialy or weekly basis. I found out about the :
http://www.boost.org/development/running_regression_tests.html
page but I wanted to know if there is any speficic need in term of paltofrm/compiler coverage or if I can just run whatever and push the regression result up ?
Joel, if you would like to dedicate some of those resources to testing for ryppl it would be extremely useful. Thanks, -- David Abrahams BoostPro Computing http://boostpro.com
On Jun 30, 2010, at 4:02 AM, joel falcou wrote:
David Abrahams wrote:
Joel, if you would like to dedicate some of those resources to testing for ryppl it would be extremely useful.
Well ^ ^ why not ;) If you can point me to the proper docs and stuff .
We will be sure to do that absolutely as quickly as we can. Ryppl development is ramping up quickly in the next week. (http://github.com/ryppl/ryppl/issues/#issue/18) -- David Abrahams BoostPro Computing http://boostpro.com
Joel Falcou wrote:
page but I wanted to know if there is any speficic need in term of paltofrm/compiler coverage or if I can just run whatever and push the regression result up ?
The release branch is still missing regular gcc-4.5 runs. The only gcc-4.5 runs in the table are from end of may, so nobody has yet setup a client to regularly run gcc-4.5: http://www.boost.org/development/tests/release/developer/summary.html I guess it would be quite useful to have regular runs of this bleeding edge compiler in both C++0x and C++03 mode also for the release branch. Regards, Thomas
Thomas Klimpel wrote:
Joel Falcou wrote:
page but I wanted to know if there is any speficic need in term of paltofrm/compiler coverage or if I can just run whatever and push the regression result up ?
The release branch is still missing regular gcc-4.5 runs. The only gcc-4.5 runs in the table are from end of may, so nobody has yet setup a client to regularly run gcc-4.5:
http://www.boost.org/development/tests/release/developer/summary.html
I guess it would be quite useful to have regular runs of this bleeding edge compiler in both C++0x and C++03 mode also for the release branch.
How can I install 4.5 in a human-bearable way ? If i got some pointer on that, I'll put it on each of the machines then.
On what kind of OS do you want to install it? Oh and by the way, is everyone free to run and upload regression tests? (I was thinking of running regressions tests on one of my setup)
On Wed, Jun 30, 2010 at 1:20 PM, Mathieu - <ptr.jetable@gmail.com> wrote:
On what kind of OS do you want to install it?
Oh and by the way, is everyone free to run and upload regression tests? (I was thinking of running regressions tests on one of my setup)
I do have a computer that is idle most of the time as well. A core 2 quad with 8GHz. It runs linux. I can run regression tests daily with two cores. I have gcc 4.4.1 x64 preinstalled but I can install another compiler. -- Felipe Magno de Almeida
On Wed, Jun 30, 2010 at 1:30 PM, Felipe Magno de Almeida <felipe.m.almeida@gmail.com> wrote:
[snip]
I do have a computer that is idle most of the time as well. A core 2 quad with 8GHz. It runs linux.
Sorry, 8GB, not 8GHz, that would be really fast! (And hot I presume) [snip] -- Felipe Magno de Almeida
On Jun 30, 2010, at 12:30 PM, Felipe Magno de Almeida wrote:
I do have a computer that is idle most of the time as well. A core 2 quad with 8GHz.
Insanely fast for COTS hardware. Are you sure you don't mean 8GB? -- David Abrahams BoostPro Computing http://boostpro.com
On 30/06/2010 17:20, Mathieu - wrote:
On what kind of OS do you want to install it?
Oh and by the way, is everyone free to run and upload regression tests? (I was thinking of running regressions tests on one of my setup)
Yep. :) Follow the instructions at <http://www.boost.org/development/running_regression_tests.html>. The result of the tests will be automatically uploaded and then displayed. KTC p.s. Where it say 5GB, think more along the line of 25+GB.
On 07/01/2010 03:15 AM, KTC wrote:
Yep. :)
Follow the instructions at <http://www.boost.org/development/running_regression_tests.html>. The result of the tests will be automatically uploaded and then displayed.
KTC
p.s. Where it say 5GB, think more along the line of 25+GB.
Hi, I just ran the tests on release branch with Linux 64 bits and g++ 4.5.0. How often should the test script run ? Daily (or more likely, nightly :)), weekly, ... ? By the way, the --mail option seems broken. I get this error message when I use it :
$ python run.py --runner=Maxime --toolsets=gcc-4.5.0 --bjam-options='-j4' --tag='branches/release' --mail='maxime@altribe.org' # Running regressions in /home/yabo/boost_test... # Creating regression scripts at /home/yabo/boost_test/tools_regression_src... # Dowloading regression scripts from http://svn.boost.org/svn/boost/trunk/tools/regression/src... # Sending start notification to "maxime@altribe.org" # Sending report to "maxime@altribe.org" Traceback (most recent call last): File "run.py", line 60, in <module> runner(root) File "/home/yabo/boost_test/tools_regression_src/regression.py", line 205, in __init__ self.main() File "/home/yabo/boost_test/tools_regression_src/regression.py", line 566, in main getattr(self,action_m)() File "/home/yabo/boost_test/tools_regression_src/regression.py", line 544, in command_regression traceback_ = '\n'.join( apply( traceback.format_exception, sys.exc_info() ) ) NameError: global name 'traceback' is not defined
If I remove it, everything goes fine. -- Maxime
On Jul 1, 2010, at 5:20 AM, Maxime van Noppen wrote:
Hi,
I just ran the tests on release branch with Linux 64 bits and g++ 4.5.0.
How often should the test script run ? Daily (or more likely, nightly :)), weekly, ... ?
More often == more better -- David Abrahams BoostPro Computing http://boostpro.com
On 7/1/2010 4:20 AM, Maxime van Noppen wrote:
On 07/01/2010 03:15 AM, KTC wrote:
Yep. :)
Follow the instructions at <http://www.boost.org/development/running_regression_tests.html>. The result of the tests will be automatically uploaded and then displayed.
KTC
p.s. Where it say 5GB, think more along the line of 25+GB.
Hi,
I just ran the tests on release branch with Linux 64 bits and g++ 4.5.0.
How often should the test script run ? Daily (or more likely, nightly :)), weekly, ... ?
Now that you are doing testing.. Please take a moment to subscribe to the testing list. -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org (msn) - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim,yahoo,skype,efnet,gmail
Rene Rivera wrote:
Now that you are doing testing.. Please take a moment to subscribe to the testing list.
I just did too ;) I ran the script this afternoon and everything went OK for trunk testing. I'll wait for the test to become visible to see if nothing crapped out and put some nightly cronjob for the linux machine. I'll setup the same thing for other tests and on the other machine asap. Tests runner name should be something like lri.upsud.linux.gcc.
Joel Falcou wrote:
How can I install 4.5 in a human-bearable way ? If i got some pointer on that, I'll put it on each of the machines then.
As it seems some others also volunteered to give test resources, I hope that somebody else will donate a gcc-4.5 test runner for the release branch. Because you have access to icc compilers, regular runs of your latest installed icc version for the release branch would perhaps make even better use of your resources, as these compilers are already installed on your machines (and it's easier for other people to get access to gcc-4.5 than to get access to icc). I use a gcc-4.5 for windows from <http://www.equation.com/servlet/equation.cmd?fa=fortran> for my tests, so I haven't built a gcc-4.5 on linux from sources myself. However, one of my colleagues has built gcc-4.5 from sources, and I haven't heard him complain, so I guess it won't be too difficult. Regards, Thomas
Thomas Klimpel wrote:
As it seems some others also volunteered to give test resources, I hope that somebody else will donate a gcc-4.5 test runner for the release branch. Because you have access to icc compilers, regular runs of your latest installed icc version for the release branch would perhaps make even better use of your resources, as these compilers are already installed on your machines (and it's easier for other people to get access to gcc-4.5 than to get access to icc).
Well, I will personnaly need 4.5 , so two-birds-one-stoning this could be a good idea too
On Jun 30, 2010, at 12:07 AM, Joel Falcou wrote:
... but I wanted to know if there is any speficic need in term of paltofrm/compiler coverage?
I too can offer up some resources for release testing. Would the Sandia trunk testers be useful for the release branch? -- Noel
participants (10)
-
Belcourt, Kenneth -
David Abrahams -
Felipe Magno de Almeida -
joel falcou -
Joel Falcou
-
KTC -
Mathieu - -
Maxime van Noppen -
Rene Rivera -
Thomas Klimpel