Hi Peter, * Peter O'Gorman wrote on Mon, Jun 20, 2005 at 04:34:00PM CEST: > Hi, > Well, Apple is switching to Intel. I figure that this means more people > will want to build fat during the transition. > > This patch (one for HEAD, one for branch-1-5) will allow -arch flags > through to the linker.
I would like some general rule to allow such stuff through. This seems rather obviously ok to me, while we also need to do something about -nostdlib and similar not-so-obvious stuff. > Okay? Minor nit below (HEAD/branch-2-0). Regards, Ralf > 2005-06-20 Peter O'Gorman <[EMAIL PROTECTED]> > > * conig/ltmain.m4sh [darwin]: Accept -arch flag. > > from Ralf Wildenhues <[EMAIL PROTECTED]> > Index: config/ltmain.m4sh > =================================================================== > RCS file: /cvsroot/libtool/libtool/config/ltmain.m4sh,v > retrieving revision 1.68 > diff -u -3 -p -u -r1.68 ltmain.m4sh > --- config/ltmain.m4sh 5 Jun 2005 17:34:17 -0000 1.68 > +++ config/ltmain.m4sh 20 Jun 2005 14:17:33 -0000 > @@ -2325,6 +2325,15 @@ func_mode_link () > prev= > continue > ;; > + > + arch) > + prev= > + compiler_flags="$compiler_flags $arg" > + compile_command="$compile_command $arg" > + finalize_command="$finalize_command $arg" > + continue > + ;; > + > framework) > case $host in > *-*-darwin*) > @@ -2559,7 +2568,13 @@ func_mode_link () > fi > continue > ;; > - > + -arch) > + prev=arch Why not prev=xcompiler and just skip the first hunk? You could even merge with the case that handles -model.. The only reason against that would have been alphabetic sorting of the switches which you just ignored anyway. :) Maybe we should just merge all of the alike-handled switches.. > + compiler_flags="$compiler_flags $arg" > + compile_command="$compile_command $arg" > + finalize_command="$finalize_command $arg" > + continue > + ;; > -framework) > prev=framework > continue
