[Boost-testing] Martin Wille -- another issue

From: Vladimir Prus (ghost_at_[hidden])
Date: 2007-01-25 10:05:59

Martin,
something seems wrong with:

        http://tinyurl.com/29xke3

there are two results. The failing one has very weird directory
name and, further, the test report itself has:

        Compile [2006-05-26 11:40:13 UTC]:

Which is hardly relevant to anything. I must admit I don't know
what's going on, and I think the only way to fix that is
in small steps.

Can you do the following:

1. Delete *all* build results.

2. Delete your results from FTP

3. Modify your regresion runner script to:
        
        - Always drag regression.py to revision 1.63.2.6, as
        we've established yesterday that later version 
        pass --toolset and break versions.

        - Run just the gcc-3.4.5_linux_x86_64 tests, to
        minimize turnaround time.

4. Start another test run.

That way we'll be able to make sure all the testing/reporting
pipeline itself works OK where other variables are known good.

Thanks in advance,
Volodya