Re: [boost] Comeau MT regressions affecting Spirit

From: Aleksey Gurtovoy (agurtovoy_at_[hidden])
Date: 2004-01-25 19:57:14

Joel de Guzman wrote:
> Giovanni Bajo wrote:
>
> > Hello,
> >
> > looking at:
> >
http://boost.sf.net/regression-logs/cs-win32_metacomm/developer_result_page.html#spirit
> >
> > most of Spirit regressions with Comeau come from Boost.Thread failures.
> > Spirit's code already handles BOOST_HAS_THREADS correctly (skip the test
if
> > there is no threading support) but this doesn't seem enough for Comeau.
The
> > output is totally cryptic to me:
> >
> > ------------------------
> > Lib  output:
> >
bin/boost/libs/thread/build/boost_thread.dll/meta-como-win32-4.3.3-vc7/debug
> > /threading-multi
> > ------------------------
> >
> > Now, why is the test reported to fail without even running it?

Basically the above means that the test failed due to the listed (and
failed) dependency ('boost_thread.dll'). Hopefully the output is less
cryptic now -- http://tinyurl.com/2h8bz.

> > How are we
> > supposed to handle this? Should we mark it as "expected-to-fail" on
Comeau?

IMO it would be totally adequate; or you can eliminate the tests' dependency
on Boost.Threads when threading is not available.

--
Aleksey Gurtovoy
MetaCommunications Engineering