Re: [boost] Beta posted for new regression tests

From: David Abrahams (david.abrahams_at_[hidden])
Date: 2002-07-23 12:43:14

Hi Beman,

What are we going to do about the Python tests?
I think they should be included in official test runs, at least.
Ralf is already doing some nightly builds at http://cci.lbl.gov/boost/,
FWIW.

-Dave

----- Original Message -----
From: "Beman Dawes" <bdawes_at_[hidden]>
To: <boost_at_[hidden]>; <boost_at_[hidden]>
Sent: Tuesday, July 23, 2002 1:32 PM
Subject: RE: [boost] Beta posted for new regression tests

> At 12:17 PM 7/23/2002, Jeff Garland wrote:
>
>  >I wonder then if it would be of benefit to generate one .html file for
> each
>  >.xml file instead of combining them?  I noticed when I
>  >hit one of the 'Fail' links and then returned to summary page that all
of
>
>  >the 'Fail' links where colored as if I had visited each
>  >one.  It would be nice if only the one I actually navigated to was
>  >highlighted.  Since web browsers only track visits on the file
>  >level this would require one html file per toolset per test.
>
> I considered that, but the management of all those files would be a
serious
> problem in our current environment.  Moving them across the Internet to a
> server is a problem.  If we had control over the server, we could
probably
> work something out, but that is hard to do as it stands now.
>
>  >I also agree with Dave that an overall summary page by library would be
a
>
>  >really nice.  The page is already long and as boost
>  >grows and adds more tests this is going to get out of hand.  I'm
thinking
>
>  >something like:
>  > Summary Results for 2002-07-23 15:24:00 for Win32 -- 3 toolsets:
>  >        * GCC 3.1
>  >        * Borland
>  >        * Metroworks
>  >
>  >  library     #tests    failures   warn   missing
>  >  filesystem    3           0        1       0
>  >  thread        1           0        1       0
>  >  tokenizer     5           2        0       0
>  >
>  >Library names would then hyperlink to an html file which contains only
> the
>  >results for that library.
>
> Yes, although we'll have to minimize the number of files unless we can
> figure out how to manage large numbers of files.
>
>  >Finally, from the library author perspective is there a reference for
> what
>  >we need to do to get our tests plugged into the system?
>
> Basically, just add your tests to status/Jamfile.  It is fairly
> self-explanatory.
>
>  >Great work!
>
> Thanks!
>
> --Beman
>
> PS: I've just refreshed the files, in the process of testing the script
to
> be run automatically.  Changed to reporting only tests that didn't pass
on
> all compilers.  Still at http://www.esva.net/~beman/jam_regr.html
>
>
> _______________________________________________
> Unsubscribe & other changes:
http://lists.boost.org/mailman/listinfo.cgi/boost