Re: [Boost-testing] [iostreams] need help understanding regression report

From: Martin Wille (mw8329_at_[hidden])
Date: 2005-02-11 15:59:12

Jonathan Turkanis wrote:
> Hi All,
> 
> I'm trying to understand the relationship between the results posted under
> "Linux: Martin Wille" here:
> 
> http://tinyurl.com/3rnxo
> 
> and the results for iostreams which are displayed here:
> 
> http://boost.sourceforge.net/regression-logs/cs-Linux.html
> 
> In particular, the platforms look exactly the same, but while the former shows
> many tests failing, the latter shows all but three passing. Furthermore, the
> former is dated yesterday but the results seem to reflect the state of CVS
> several days ago.
> 
> Finally, the 'unresolved issues' for iostreams:
> 
> http://www.meta-comm.com/engineering/boost-regression/developer/issues.html
> 
> is inconsistent with both of the above.
> 
> I'm sure there are simple answers to these questions, but I have only recently
> begun examining these tables in detail and don't know what to make of them.

The table shown under
http://boost.sourceforge.net/regression-logs/cs-Linux.html
is the most recent one. Meta-comm needs some time to integrate
the results into their version of the tables. Occasionally,
that processing fails for whatever reasons. That causes the
tables shown at the meta-comm site sometimes to be out of
date. Frankly, I have the impression the XML based processing
is a little fragile and I decided to continue uploading the
old-fashioned format of the result in addition to the new
XML based format.

HTH,
m