Re: [Boost-testing] [EXTERNAL] Regression zip failure

Subject: Re: [Boost-testing] [EXTERNAL] Regression zip failure
From: Beman Dawes (bdawes_at_[hidden])
Date: 2014-01-30 10:11:32

On Wed, Jan 29, 2014 at 11:30 PM, Belcourt, Kenneth <kbelco_at_[hidden]>wrote:

>
> On Jan 29, 2014, at 5:57 PM, Belcourt, Kenneth wrote:
>
> > On Jan 29, 2014, at 4:08 PM, Steven Watanabe wrote:
> ...
>
>> FWIW, the reporting system trims the output
> >> to 65536 bytes.
> >
> > Good to know.  I was thinking we could constrain bjam to match the
> reporting system (-m64) so I just built with clang on develop and the files
> are order 100 Mb with -m64.
> >
> > -rw------- 1 kbelco kbelco 129370345 Jan 29 17:52 bjam.log
> > -rw------- 1 kbelco kbelco 100987295 Jan 29 17:54 Sandia-Linux.xml
> >
> > Not too bad so for now I'll limit bjam targets to 64kb, I'll get a patch
> in for this shortly.
>
> Updated develop regression.py to add -m64 to bjam options if user didn't
> pass a -m option (will use users -m option if they did).  This will impact
> both develop and master testers since they both use the develop tools.
>

Good, for now.  You might even try a much smaller number, and see if anyone
complains:-)

Long term, it would be great to be able to optionally report
pass/fail/warn. Possibly run two versions of the reports, one with the
--warn option and one without. Then both developers and users could spot
problems earlier than we do now.

--Beman