Re: [Boost-testing] Failing to run regression tests

From: Roland Schwarz (roland.schwarz_at_[hidden])
Date: 2006-10-20 12:42:19

Rene Rivera wrote:
> It's just a matter of adding "<build>no" as a requirement at the 
> appropriate times.

I recognized this, but there ought to be a way to depend on availability
of some other library which might not be able to build.

My fix does this, but making undocumented use of some rules that
belong to spirit. I have seen no simple way to test if another
library is buildable, and emit <build>no if not.

Of course it also would be possible to track the toolset directly,
but this does not exactly express intention. E.g. when later on
the dependant library is fixed the build does not automatically
promote to depending libs.

Roland