On Sunday 26 November 2006 02:10, Rene Rivera wrote:
> Vladimir Prus wrote:
> > Hi,
> > the gcc toolset accepts the 'flavour' option, but
does not make any use
> > of it. Why is it needed?
>
> Ah, what do you mean it doesn't use it? ... lines
90-103:
>
> local condition ;
> if $(flavor)
> {
> condition = [ common.check-init-parameters gcc
>
> : version $(version)
> : flavor $(flavor)
>
> ] ;
> }
> else
> {
> condition = [ common.check-init-parameters gcc
>
> : version $(version)
>
> ] ;
> }
I saw that, but that does not affect anything except for
target directories
and the full name of the toolset -- and does not effect any
command lines at
all.
> > Also, it accepts the 'archiver' option, and has
this comment:
> >
> > #~ If it's not a system gcc install we should
adjust the various
> > #~ programs as needed to prefer using the
install specific versions.
> > #~ This is essential for correct use of MinGW
and for
> > cross-compiling.
> >
> > This is obviously important if there's no system
'ar'. However, is this
> > important when there is system 'ar' already? I
don't think the 'ar' tool
> > does any thing that is target specific. Am I
wrong?
>
> I don't know if you are wrong or not :- But it makes
sense that if a
> version specific install of gcc has its own 'ar' then
the person
> installing it means for it to be used regardless of
whether it does
> anything different than the system 'ar'. More
technically 'ar' is meant
> to produce files that the linker can consume. If the
linker is specific
> to the install, for example for cross-compilers, then
it's essential
> that it use the cross-compile specific version as it
*could* do
> something specific for the target platform.
Hmm, I guess I can put "specify this just in case"
in docs, but it still does
not feel right 
- Volodya
_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost-build
|