This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
AR_FOR_TARGET (Was: Re: PATCH: look for host ranlib in configure)
Alexandre Oliva wrote:
> > Also, in my next patch I'm trying to move {RANLIB|NM|AR}_FOR_TARGET
> > macros to configure too. Will it need similar treatment?
>
> I believe so. In the top level, you should probably arrange for them
> to be substed with $(AR), $(NM) and $(RANLIB) (as opposed to the
> contents of $(AR), $(NM), etc) in case host_canonical =
> target_canonical, and with the values they're regularly replaced with
> otherwise. But this won't save us from duplicating the logic within
> GCC's configure.in, since it should not depend on variables passed
> from the top level. Or should it?
I see that AR_FOR_TARGET etc. is passed to all subdirs. If we remove those
definitions from top level Makefile, most probably target libraries will
break in cross compilation. And if we don't, we can simply remove all those
AR_FOR_TARGET definitions in gcc/Makefile.in - they are overriden by top
level Makefile _all the time_. Any comments?
Laurynas