Thanks, I just committed the fix. Haven't been testing gcj much, so your input is very helpful!
Stan Andrew Pinski wrote: > While compiling gcc with java enabled (but not libjava), the fsf way, > I found that there was an > error in the source code where flag_altivec was defined in java/lang.c > which is wrong now that > the variable is also defined in toplev.c which gets included when > linking by libbackend.a > > Thanks, > Andrew Pinski > > PS here is a patch for java/lang.c, which just comments out the > variable, bootstrap in progress: > > Index: lang.c > =================================================================== > RCS file: /cvs/Darwin/gcc3/gcc/java/lang.c,v > retrieving revision 1.22 > diff -u -r1.22 lang.c > --- lang.c 26 Aug 2002 05:41:23 -0000 1.22 > +++ lang.c 29 Aug 2002 03:58:53 -0000 > @@ -917,9 +917,13 @@ > return -1; > } > > +/* ATP LOCAL begin APPLE */ > +/* no need for flag_altivec anymore */ > /* APPLE LOCAL begin AltiVec */ > /* Placeholders to make linking work, remove when altivec support is > correct */ > -int flag_altivec; > +/*int flag_altivec;*/ > +/* APPLE LOCAL end AltiVec */ > +/* ATP LOCAL end APPLE */ > > int > comptypes (type1, type2) > >
