Boost logo

Boost-Build :

From: Vladimir Prus (ghost_at_[hidden])
Date: 2003-10-15 08:34:30


Austin Bingham wrote:
> > The problem is that "use-project" is allowed in Jamfile, too, and I think
> > it's a good idea. It's still possible to make "use-project" work, anyway.
> > I was wondering about use-cases for "use-project" where used project does
> > not have a separate project-root.jam. In my settings, I use "use-project"
> > for dependedies on boost, and some local, relatively standalone
> > libraries, which all have their own project-root.jam. So I wonder if
> > other case is reasonable/ needed.
>
> I used use-project fairly extensively to point to targets with the same
> project-root to give flexibility when I was organizing my source files. It
> gave me the freedom to move things around freely while minimizing the
> ripple of changes. So, I guess that's one use case. It could also be
> argued that, in the case of particularly deep project hierarchies,
> use-project provides the best way to give make building those targets easy
> (i.e. by giving them easy names).

Ok, you've convinced me.

> > Of course, it is possible to delay processing of "use-project" directives
> > until
>
> Yes, that was sort of what I had in mind. Whatever module does the Jamfile
> loading could always remember what it has already loaded...if a
> use-project directive asks for it to be loaded again, it could just be
> ignored. I worry that I'm missing some subtlety about Jamfile load order,
> though.

Actually, the problem was fairly technical. The project.load rule first marks
the module as loaded, and then calls other rule. That rule loads parent
Jamfile, which contains "use-project" and tries to load the first Jamfile.
The problem is: it's marked as loaded, so it's not loaded again. But it was
never read, so project id for that module is not available, and use-project
crashes.

I guess I'll fix this by marking Jamfile as loaded somewhat deeper in the
process.

- Volodya

 


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