Re: [boost] Regression tests on more platforms?

From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-01-17 17:16:37

At 10:27 PM 1/17/2001 +0100, Jens Maurer wrote:

 >Beman Dawes wrote:
 >> "status" sounds good to me.  I'm tied up trying to get out a release of
 >> some non-boost software.
 >
 >:-)
 >
 >>   If you want to move things over, that is fine
 >> with me. That will fix the screwed-up cs-win32.zip, too.
 >
 >Done.  I hope I fixed all HTML references.

Thanks.  Don't worry about the HTML references; the process I use to update 
the web site catches broken intra-site links, and can do global fixes.

 >We should reconsider whether and how we want to check in the compile
 >log files.
 >Binary files are not storage efficient with CVS's "save the latest
 >version in full and incremental text diff's for the older versions".
 >For each checked-in binary file, CVS basically stores each version
 >in full.  10 check-in's of 20 KB is 200 KB already, and we expect to
 >run the regression tests rather often.  On the other hand, the
 >non-compressed log files are 200-500 KB.
 >
 >Is there possibly some non-version controlled place we could put
 >the log files and add a link?  The log files are only really
 >interesting for developers and not for boost users, because the
 >error messages of the compilers usually require an advanced
 >understanding of the inner workings of the boost libraries.

SourceForge has an FTP facility we are not using now.  How about 
that?  (But I don't know if it is possible to link to it.)  Otherwise we 
can set up a dummy eGroups mailing list with the developers having write 
access, with the Files section open to the public.

--Beman