Re: [Boost-testing] [BoostConsulting][numeric/interval] misleading toolset name

From: Joaquin M Lopez Muñoz (joaquin_at_[hidden])
Date: 2007-02-03 08:32:28

David Abrahams <dave <at> boost-consulting.com> writes:

> 
> Joaquín Mª López Muñoz <joaquin <at> tid.es> writes:
> 
> > [...]these tests use a 64bit version of MSVC 8.0 on a 64bit
> > platform that's not supported by Boost.Interval.
> >
> > The problem is that the toolset name used is "msvc-8.0", which is
> > reserved for the "standard" 32bit version,
> 
> Just to satisfy my curiosity, when/where was that decided and/or
> documented?

It's not a explicit stated declaration (hence my quoting
the word "standard"), but it is implied by the following:

* Regression results for 1.33.1 include a set of results labeled
"vc-8_0" which were produced by a 32 bit version of MSVC 8.0
(see http://engineering.meta-comm.com/boost-regression/
CVS-RC_1_33_0/RudbekAssociates.html )

* These regression results have been transferred to boost/
libs/expected_results.xml, where they're being used to detemine
what's to be considered a regression in 1.34 and what's
a new failure.

* expected_results.xml marks "msvc-8.0" as an alias of "vc-8_0".

Joaquín M López Muñoz
Telefónica, Investigación y Desarrollo