|
|
Boost Testing : |
From: Caleb Epstein (caleb.epstein_at_[hidden])
Date: 2005-03-12 00:26:05
On Fri, 11 Mar 2005 19:58:33 -0700, Jeff Garland
<jeff_at_[hidden]> wrote:
> On Fri, 11 Mar 2005 12:00:45 -0500, Caleb Epstein wrote
> > On Fri, 11 Mar 2005 11:56:32 -0500, Caleb Epstein
> > <caleb.epstein_at_[hidden]> wrote:
> >
> > > Looks like *all* of the testing results disappeared
> > >
> (http://www.meta-comm.com/engineering/boost-regression/cvs-head/developer/summary.html
> > > shows "Page not Found!") sometime after I sent my email. Perhaps they
> > > are being regenerated as I write this.
> >
> > OK, the results are back again. The tinyurl links I sent originally
> > should work again.
>
> And now that I'm back from making a living...
>
> >My copy of testtz_database.cpp is dated March 6 and appears to be
> >up-to-date w/r/t version 1.4 of that file in CVS.
>
> The results on the website were generated with the old version. We can tell
> because the output text in the new version doesn't match the output in the
> test results. Basically, you should be seeing "Caught Bad field count
> exception" fail before the program aborts if you were testing with 1.4. So,
> your code may be up to date and it wasn't recompiled or the results weren't
> processed -- don't know.
I think the CVS tarballs were wedged for a few days, which might
explain that file from 3/6 showing up only today on my system despite
its timestamp. I concur the results on meta-comm were probably
generated on the prior day from the older code. Should clear up
tomorrow (I think I set the cron job to run every day not just
weekdays).
> ps: should I just mark SunPro as "hopeless"? It looks like some of the basic
> issues have to with the compiler not handling boost::operators. It would take
> alot of work to extract that code...
Its pretty hopeless. The version I have is 5.3, which fails to
compile great swaths of Boost (though a number of useful pieces *do*
work). It may make sense to stop running the tests with this
tool-chain. Perhaps someone else out there in Boostville has a newer
SunPRO (5.5 or later I believe) and wouldn't mind running the
regression tests?
-- Caleb Epstein caleb dot epstein at gmail dot com