Re: [Boost-testing] Help with boost testing required

From: Robert Ramey (ramey_at_[hidden])
Date: 2007-09-04 11:45:53

Vladimir Prus wrote:
> David Abrahams wrote:
>
>>
>> on Mon Sep 03 2007, Vladimir Prus <ghost-AT-cs.msu.su> wrote:
>>

First of thanks to all who are helping me out with this.

A short recap of how I got here.

I wanted to be able to produce a complete table of test results for all
variants for any or all libraries.  This entails some relatively minor
adjustments to compiler_status.cpp.  I've called this library_status.cpp.

On testing this, I found that xml files produced by process_jam_log
contained "test_type=" and I needed this to be filled in in order to
produce the output I wanted.  So I made a couple of tweaks
so that this would be filled in.  I tested my Library_status script
which ran bjam, process_jam_log and library_status and was
quite satisfied.  So I checked in my changes.

It became apparent that in the course of making my changes
I made an error which was only apparent to the regression
test process.  My xml contained "library=" rather than
say "library=serialization".  Damn.  Rene rolled back my
changes and I made some more corrections.  Now I want
to try again - but having messed up once, I wanted to run
a test on my own machine.  I looked at the regression.py
script and after a little experimentation concluded that it
would be easiest just to follow the instructions at
http://www.boost.org/tools/regression/xsl_reports/runner/instructions.html

Its apparent that it should just download the tarball,
unpack it, run  the tests - with my own process_jam_log
and post the results.  Cool - just do it.  After all, regression
testers are doing with no problem.

But it didn't work for me.  It seems that the tarball is missing
something that is required.  I'm reluctant to spend time
on this because - a) its suppose to work out of the box
and b) this is sort of a side issue to other stuff I'm trying
to get done.

Really, I read the instructions carefully, checked that I had
the correct version of python, etc and invoked the commands
required.  So either:

a) I missed something obvious in the instructions - err - I've done this 
before.
b) Something is missing in the instructions - could easily happen
c) Something is broken - also a possibility.

So that's how I got here.  A torturous road considering all I
really want to do is to test my latest change in process_jam_log.

Robert Ramey