This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Failed compile, SCO OSR 5.0.5
- To: Jean-Pierre Radley <jpr at jpr dot com>, EGCS Developers <egcs at cygnus dot com>
- Subject: Re: Failed compile, SCO OSR 5.0.5
- From: Robert Lipe <robertl at dgii dot com>
- Date: Wed, 9 Sep 1998 10:18:49 -0500
- References: <19980908155644.I14075@jpr.com>
> cc -c -DIN_GCC -g -I. -I.. -I../../../egcs/gcc/java -I../../../egcs/gcc/java/.. -I../../../egcs/gcc/java/../config -I../../../egcs/gcc/java/../../include ../../../egcs/gcc/java/mangle.c
> "../../../egcs/gcc/java/javaop.h", line 54: error: invalid type combination
> "../../../egcs/gcc/java/javaop.h", line 116: warning: shift count negative or too big: << 32
> "../../../egcs/gcc/java/javaop.h", line 116: warning: shift count negative or too big: << 32
Hmmm. I don't know why I didn't notice this yesterday.
I think this will fail for any bootstrap compiler that doesn't support
'long long'. I suspect the real problem is that java shouldn't be built
in stage1 during a bootstrap and should have the build deferred until
the stage2 (xgcc) compile is availble. Why is it being built at all?
$ make LANGUAGES=c
Makefile:1862: warning: overriding commands for target `chill'
Makefile:868: warning: ignoring old commands for target `chill'
Makefile:1885: warning: overriding commands for target `chill-cross'
Makefile:891: warning: ignoring old commands for target `chill-cross'
Makefile:1890: warning: overriding commands for target `cc1chill'
Makefile:896: warning: ignoring old commands for target `cc1chill'
[ bunch of similar chill warning ]
[ bunch of tests to see if we might eventually build g77 ]
cd java && make "AR_FLAGS=rc" "AR_FOR_TARGET=` if [ -f /play/tmp/n/gcc/../binutils/ar ] ; then echo /play/tmp/n/gcc/../binutils/ar ; else if [ "i686-pc-sco3.2v5.0.5" = "i686-pc-sco3.2v5.0.5" ] ; then echo ar; else t='s,x,x,'; echo ar | sed -e $t ; fi; fi`" "BISON=`if [ -f /play/tmp/n/gcc/../bison/bison ] ; then case /p[ billion args passed to java build ]
RJL