From: David Abrahams (david.abrahams_at_[hidden])
Date: 2002-02-28 12:09:50
----- Original Message -----
From: "Rene Rivera" <grafik666_at_[hidden]>
> On 2002-02-28 at 11:25 AM, david.abrahams_at_[hidden] (David Abrahams) wrote:
> >IMO Thomas is on the right track; if you really want a global requirement
> >set it would make just as much sense to use a global variable.
> I don't need a global requirement set.
Sorry, I meant "you" in general, not you specifically.
> Mostly I was thinking that
> variant/requirements/build directives should mirror each other, not just
> syntax but in semantics.
In that case I buy the idea that we have a few global variants, but that
users can define additional variants which are namespaced according to their
module/subproject by default. Likewise for requirement-sets, but I can't see
that we'd supply global requirement sets... can you?
Maybe it makes sense to use C++-style, rather than Python-style namespacing
for these things. IOW, the names would be available without qualification to
all subprojects of a given project.
> And I thought we we're trying to move away from variables in general. I
> this is going to be disposable code when V2 gets done. But I think getting
> idea of "good" practice now is also important.
It may be disposable code, but hopefully not disposable interface.
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