Re: [jamboost] Re: V1 testing question

From: Vladimir Prus (ghost_at_[hidden])
Date: 2003-11-03 04:17:53

David Abrahams wrote:
> Vladimir Prus <ghost_at_[hidden]> writes:
> > I've just made the "regression.py" test pass with msvc and borland, and I
> > hope to run the entire Boost regression suite on those compilers soon. I
> > have just one question now. V1 testing.jam contains:
> >
> > if $(gCURRENT_TOOLSET) in $(gcc-compilers)
> > {
> > on $(target)
> > {
> > RUN_PATH = [ unique $(p) $(CYGWIN_ROOT)/$(p) $(RUN_PATH)
> > $(ENV_PATH) ] ;
> > }
> > }
> >
> > What is the purpose of this code?
>
> Well, it's augmenting the list of directories that will be stuck into
> PATH before the executable is run. The reason for adding
> $(CYGWIN_ROOT)/$(p) is that if the developer of the Jamfile was using
> Cygwin bjam, the absolute path "/foo/bar" actually maps to
> "$(CYGWIN_ROOT)/foo/bar" for the person running the tests under
> regular windows bjam. 

I'm afraid I don't understand this. Most of the the time 'p' is implicitly 
generated by Boost.Build, so under regular windows bjam it will still be 
correct. nt

> We're also tacking on $(ENV_PATH) so the
> user's PATH environment variable may be considered.

Yea, that's done in other cases and I've copied that behaviour.

- Volodya