Re: [jamboost] Autolinking and Boost.Build

From: Vladimir Prus (ghost_at_[hidden])
Date: 2005-08-02 11:10:02

On Tuesday 02 August 2005 18:51, John Maddock wrote:
> >Don't regression test build just a single variant? No debug/release
> >single-threaded/multi-threaded static-runtime/dynamic-runtime combos. If
> >there are no all such combos, then the testing of autolink is not so
> >comprehensive.
>
> True, but see below.
>
> >Answering your question -- it might be possible to add BOOST_ALL_NO_LIB
> >conditionally.
>
> Umm, what good does that do?
>
> OK, lets see if we can design a descent testing suite, just thinking out
> loud here:
>
> 1) Can I have a stage target that directs the "installed" lib to a path of
> my choice?

Yes:

install your_stage : your_lib : <location>your_location ;

> 2) Can I make the build of an executable depend upon a stage target being
> built first? (Last time I asked for bbv1 the answer here was no).

Yes:

exe your_exe : ........ : <dependency>your_stage ;

> 3) I can add <runtime-link>static etc build requirements to a target, but
> can I make "release" or "debug" also build requirements?

The following should work:

exe your_exe : ............ : <runtime-link>static <variant>release ;

> If "yes" to all of the above I think we could do something like define one
> exe and one lib target for each build combination (tedious but there are
> only eight of them, so it's do-able), and then all the combinations would
> (hopefully) get tested?

Looks so. Would you like to to sketch an implementation of this for some 
library?

- Volodya

-- 
Vladimir Prus
http://vladimir_prus.blogspot.com
Boost.Build V2: http://boost.org/boost-build2