|
|
Boost Testing : |
From: Roland Schwarz (roland.schwarz_at_[hidden])
Date: 2007-01-18 12:12:19
Aleksey Gurtovoy wrote:
> Roland Schwarz writes:
>> I doubt that the regression.py really is very failsafe,
>
> Meaning?
Thank you for your answer. I got a clearer picture now. I was not aware
that all of the issues I have seen are already are being worked on.
To be honest, I realized from looking at the regression results that
they currently do not correspond from what I get when doing manual test
runs. Erreonously I attributed this to regression.py.
Then Anthony Williams told me that he got different results when doing a
manual run than with regression.py (don't know if he found out whats
going wrong).
Then I ran into the problem with the message boxes. Yes this partly is
my fault, since I did not see the explanation to --monitored under
http://boost.org/tools/regression/xsl_reports/runner/instructions.html
(Which btw. had an incorect ling to the download site. I corrected this
in the CVS.)
Another issue was that while trying to test for msvc-6.5 the run failed
since the build of process_jam_log is not feasible with this toolset. It
took me quite some while until I came upon the --pjl-toolset switch. I
had to guess its meaning. First I was thinking that it takes a path to a
pre-puilt until I found by try and error that it is a compiler name.
> short, the above is not true.
Ok, thank you for having explained. I think I understand better now.
Roland