Fwd: Re: Boost.Threads and Regression Testing

From: bill_kempf (williamkempf_at_[hidden])
Date: 2002-02-28 10:17:46

--- In jamboost_at_y..., "David Abrahams" <david.abrahams_at_r...> wrote:
> 
> ----- Original Message -----
> From: "bill_kempf" <williamkempf_at_h...>
> 
> > Yes. Now I guess the point (beyond the link bug in the build 
system
> > for <dll>)
> 
> OK I fixed that one. Turns out it was mostly my fault.

Thanks! Being part of the regression testing system will help out a 
lot with this library. Especially since I have access to so few 
compilers/platforms.

> Your test seems to be crashing with CWPro7 and MinGW.

Crashing or failing?

> It works with msvc and
> borland. It doesn't compile with intel 5 or 6, and seems to run 
forever with
> Cygwin.
> 
> > is that I think a build section/rule variable should be
> > included in the regression tests. It's important that I be able 
to
> > insure the various subvariants compile, link, and pass runtime
> > tests.
> 
> Yeah, probably.
> 
> > This will become more important if I can turn pthread-win32
> > support into a variant on Windows platforms.
> 
> That should probably be a build property, not a variant.

As a build property it requires running the tests multiple times with 
some annoying variable specifications (and also requires cleaning, 
which means I have to run the ENTIRE regression test). As a variant 
I can compile/test both with one Jam invocation (and can use 
environment variables to eliminate the need for specifying PTW32 on 
the command line).

Thanks,

Bill Kempf