[Boost-testing] Release results... was: Branch agnostic testing...

From: Rene Rivera (grafikrobot_at_[hidden])
Date: 2007-10-16 12:13:41

Beman Dawes wrote:
> Beman Dawes wrote:

>> I'm working on the real release branch reports right now.

> It would help me if you would run the release branch reports once on 
> your machine, and then send me the script you used, the starting 
> directory structure you ran it in, and tell me the point in the 
> directory structure you ran the script from.

OK, did a run on my machine, took a while since I had to wait for a good 
stopping point in the cycle (while I was awake). I also did a some minor 
changes to the build_results.sh script so I could tell it from the 
outside which branch I was testing. My tree looks like so:

$HOME/devroots/Boost/
   run-testing.sh
   boost-regression/
     boost/
     build_results.sh
   boost-release-testing/
     run.py
   boost-trunk-testing/
     run.py

"run_testing.sh" is my master script which does the report building and 
testing. I run it from the "Boost" directory as "./run-testing.sh". For 
the report generation for the release it now does:

     echo "************************ RESULTS ***********************"
     cd $HOME/devroots/Boost/boost-regression
     rm -f build_results.sh*
     #~ wget 
http://svn.boost.org/svn/boost/trunk/tools/regression/xsl_reports/build_results.sh
     cp boost/tools/regression/xsl_reports/build_results.sh .
     nice sh ./build_results.sh release 2>&1 | tee results.log
     echo "********************************************************"
     sleep 300

I swap the "wget" vs. the "cp" based on whether I'm making changes to 
the build_results.sh script or not. As you can see I run the 
"build_results.sh" from the "Boost/boost-regression" directory. Since 
the "build_results.sh" script is copied into that, then only prep I need 
to do is have the dir structure. In particular the 
"Boost/boost-regression/boost" tree. This is a checkout of the trunk 
source tree which gets updated by "build_results.sh" on each run. But to 
make the update work faster I switched out (svn switch) all the subdir 
that are not important to the "empty" tree in the SVN repo:

     S  regression
     S  people
     S  boost
     S  libs
     S  tools/litre
     S  tools/release
     S  tools/quickbook
     S  tools/boostbook
     S  tools/wave
     S  tools/bcp
     S  tools/buildbot
     S  tools/inspect
     S  more
     S  wiki
     S  doc

The generation worked fine for me, yes I have stuff in the merged dir. 
You can see the results at 
<http://beta.boost.org/development/tests/release/developer/summary.html>. 
But I'll switch back to trunk regressions now :-)

HTH.

-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com
-- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo