Re: [Gm2] OSX, "by hand" - failure on ranlib (not first one, although)

[Gaius Mulley]

Dragiša Durić <address@hidden> writes:

> Here goes.
>
> On Jan 5, 2011, at 2:28 PM, Gaius Mulley wrote:
>
>> ...
>> file gm2/ulm-lib-gm2/sys/SYSTEM.o gm2/ulm-lib-gm2/sys/UnixString.o
>
> ~/gm2-1.0.0/build-4.1.2% cd gcc
>
> gm2-1.0.0/build-4.1.2/gcc% rm gm2/ulm-lib-gm2/sys/SYSTEM.o 
> gm2/ulm-lib-gm2/sys/UnixString.o
>
> gm2-1.0.0/build-4.1.2/gcc% make gm2/ulm-lib-gm2/sys/SYSTEM.o 
> gm2/ulm-lib-gm2/sys/UnixString.o
> gcc -c -g gm2/ulm-lib-gm2/sys/SYSTEM.c -o gm2/ulm-lib-gm2/sys/SYSTEM.o
> ./xgm2 -fsoft-check-all -g  -funbounded-by-reference -fpim -fextended-opaque 
> -Wpedantic-cast -Wpedantic-param-names -ffunction-sections -fdata-sections  
> -c -B./stage1/gm2 -B./ -I../../gcc-4.1.2+gm2-20101211/gcc/gm2/ulm-lib-gm2/sys 
> -I../../gcc-4.1.2+gm2-20101211/gcc/gm2/gm2-libs 
> ../../gcc-4.1.2+gm2-20101211/gcc/gm2/ulm-lib-gm2/sys/UnixString.mod -o 
> gm2/ulm-lib-gm2/sys/UnixString.o
> ../../gcc-4.1.2+gm2-20101211/gcc/gm2/ulm-lib-gm2/sys/UnixString.mod:0: 
> warning: -ffunction-sections may affect debugging on some targets
>
> gm2-1.0.0/build-4.1.2/gcc% file gm2/ulm-lib-gm2/sys/SYSTEM.o 
> gm2/ulm-lib-gm2/sys/UnixString.o 
> gm2/ulm-lib-gm2/sys/SYSTEM.o:     Mach-O 64-bit object x86_64
> gm2/ulm-lib-gm2/sys/UnixString.o: Mach-O object i386

Hi,

ah right - so it looks as if SYSTEM is being built by the wrong
compiler.  It should be using:

./xgcc -B./ -c -g gm2/ulm-lib-gm2/sys/SYSTEM.c -o \
gm2/ulm-lib-gm2/sys/SYSTEM.o

or alternatively the configure arguments to gcc require:
   --with-cpu=default32

I wonder if you could also try:

gcc -v

to verify that this is the case (or maybe an alternative flag forces it
to produce 32 bit code)

regards,
Gaius