Re: [boost] Better regression testing

From: David Abrahams (abrahams_at_[hidden])
Date: 2000-12-24 16:57:02

Wow, how did so many tests begin to fail with STLport 4.0? Is the test
configuration OK?
-Dave

----- Original Message -----
From: "Beman Dawes" <beman_at_[hidden]>
To: "Boost.org mailing list" <boost_at_[hidden]>
Sent: Sunday, December 24, 2000 4:47 PM
Subject: [boost] Better regression testing

> Jens Maurer has contributed a new regression test program, regression.cpp,
> to replace regrtest.py.  It has been checked into the CVS (in libs
> subdirectory.)
>
> Jens' program has several key advantages:
>
> *  It is driven by configuration files, allowing use in multiple
> contexts.  For example, there could be both the main set of boost
> regression tests, and further domain library tests.  The configuration
> files also make it much easier to add additional compilers, operating
> systems, or program tests.
>
> *  It supports multiple test types:
>
>      - compile only, should succeed
>      - compile only, should fail
>      - compile + link, should succeed
>      - compile + link, should fail
>      - compile + link + run, should succeed
>      - compile + link + run, should fail
>
> *  It is written in very standard C++, and so doesn't require Python
> availability.
>
> In addition to these major advantages, there is a reasonable chance that
> the approach can be extended to do library builds.
>
> For a sample output, see:
>
http://cvs.sourceforge.net/cgi-bin/cvsweb.cgi/~checkout~/boost/libs/cs-win32
.html?rev=1.11&content-type=text/html&cvsroot=boost
>
> Lots of thanks to Jens!  (Dave Abrahams, Beman Dawes, John Maddock, and
> Jeremy Siek also contributed ideas.)
>
> --Beman
>
>
>
>