From: Vladimir Prus (ghost_at_[hidden])
Date: 2006-10-10 03:47:09
On Sunday 08 October 2006 12:25, Bojan Resnik wrote:
> > > I also ran the pch.py test with the command lines
> > > pch.py msvc-6.0
> > > pch.py msvc-7.1
> > > pch.py msvc-8.0
> > > and they completed fine, after I patched pch.py to use #include
> > > "pch.hpp" instead of #include <pch.hpp>. The patch is also attached.
> > Can you check if changing:
> > cpp-pch pch : pch.hpp : <toolset>msvc:<source>pch.cpp ;
> > to
> > cpp-pch pch : pch.hpp : <toolset>msvc:<source>pch.cpp <include>. ;
> > instead of changing the form of includes fixes the problem?
> Indeed, this does fix the problem. Does this mean that pch still
> does not inherit the requirements of its parent target (in this case
> hello executable)?
Yes. I'm not yet sure if we should ignore the problem, or do something.
> There is still the problem with pch that sometimes the source that
> creates the precompiled header file is compiled with incorrect
> commands (please see
> Franz Schnyder suggested a patch to msvc.jam that fixes the problem,
> and I am attaching a slighltly modified version of his patch that is
> made against the current CVS head.
I'll look into that as soon as possible.
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