From: Vladimir Prus (ghost_at_[hidden])
Date: 2003-03-07 10:57:21
David Abrahams wrote:
> Vladimir Prus <ghost_at_[hidden]> writes:
> > David Abrahams wrote:
> >> That's line 583 in my copy, but there's definitely a bug in that line:
> >> c:/boost/tools/build/new/virtual-target.jam:583:
> >> toolset.set-target-variables $(self.action-name) $(actual-targets)
> >> :
> >> zero is not a valid index. Also:
> >> c:/boost/tools/build/new/property.jam:148: if $(x) <
> >> $(y)
> > Guess we should fix it too? But I don't know what the code is supposed to
> > do.
> Done. Unfortunately $(x) seems to be the same as $(x) for all
> x. I wonder if we shouldn't cause an error instead?
Sloppy rules for variable expansion was giving us some concerns a long time
ago but we never got to causing errors in this particular case.
> >> I ran some simple tests and it seemed to work out just fine: both
> >> of the targets' variables would be used to build the pair.
> > Please see the attached file. It causes settings only for one target to
> > be in effect.
> OK, I'm convinced.
OK. BTW, how tests behave with msvc now?
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk