From: Vladimir Prus (ghost_at_[hidden])
Date: 2004-07-20 07:19:34
David Abrahams wrote:
> > Ok, I get you. Though we'd still need to enumerate all libraries.
> Where? Why? How? (seriously)
Where: In top level Jamfile.
Why: to know which libraries to install. We either need the names of
directories or target ids.
How: for Boost, some simple GLOB-based approach would be enough. Not sure
about wider scope.
> >> I don't know what "auto-compute subproject ids" means, but I don't
> >> think it's needed if we have the above.
> > I meaning adding prepending boost "boost" to project-ids of all
> > libraries.
> Are you talking about the capability of using relative project ids?
I think so.
> > I think there are some questions to be solved yet:
> > 1. Should we retain ability to specify absolute ids in 'project' rule
> > invocation.
> First tell me what files will invoke "project" and why.
Jamfiles for individual projects: mostly to specify requirements and
usage-requirements which apply only to those projects.
> > What are we to do if this id has several components? E.g. will
> > "/boost/foobar", be looked up by searching global id "/boost/foobar", or
> > asking project "/boost" for subproject "/foobar".
> > 2. What's the syntax to specify id->dir mapping? Just function in
> > Jamfile?
> Do you mean subproject id->directory mapping?
> Well, a function is more flexible, but less declarative. You could
> do it declaratively with SUBST, but I guess I like the function.
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