[boost] [urgent] Testing help needed..

Subject: [boost] [urgent] Testing help needed..
From: Rene Rivera (grafikrobot_at_[hidden])
Date: 2013-12-29 15:41:54

I finally have the changes to the regression scripts sufficiently complete
that I've managed to do a few minimal cycles of regression testing from the
git repos. I now need some help in various ways:

1. General testing..

I need some of the regular testers to do runs with their setups with the
new scripts. To test I suggest starting from a new testing location. And
follow the usual instructions except get the run.py script from <
https://raw.github.com/boostorg/boost/develop/tools/regression/src/run.py>.
The options are exactly the same as the SVN version of the scripts (the
--tag is mapped to master or develop as needed). If you run patch scripts
you will have to adjust paths for the new arrangement. In particular the
boost tree is now checked out to <testing-root>/boost_root.

2. Non-nix testing..

I don't have the time or resources available to check if the scripts and
git invocations work on Windows. I need one person to test and likely make
changes to the regression script to do the right things on Windows. Testing
should be the same as above. But doing changes locally and testing them is
challenging. The procedure I use is (as shell commands):

> mkdir testing-dev
> cd testing-dev
> link <BOOST_DEVELOP>/tools/regression/src/run.py run.py
> ln -s < BOOST_DEVELOP>/tools/regression/xsl_reports xsl_reports
> ln -s < BOOST_DEVELOP>/tools/regression/src tools_regression_src
> ./run.py --runner=<RUNNER_NAME> --skip-script-download --use-local

That lets you edit the regression.py and run.py scripts in your local
develop checkout. The options make it skip most of the run.py code, but
otherwise make it behave as the regular testing except using your local
changes. If you need to test changes to run.py in the normal use case you
will have to commit+push the changes and then test.

3. Report generation..

It seems that the reports are failing to get generated for some reason. I
suspect that the new scripts putting the git commit SHAs as the tested
version number are messing up the results generation. I need someone
familiar with the result generation to fix this.

-- 
-- 
-- 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