🚀 go-pugleaf

In article <1992Sep4.102238.1@cc.curtin.edu.au> smackinl01@cc.curtin.edu.au writes:
>Bill Pearson's problems may be related to the command line switches he was
>using. From memory, his problem was that "GCC filename" would fail. With the
>non-emx port, you need to use instead "GCC -c filename.c" to produce a x86
>specific object file instead of the unix .o format. GCCv2.2.1 seems to have
>difficulties creating object files in the .o format, so perhaps someone could
>clarify this?

GCC/2 creates .o object code but immediately converts it to .obj using
o2obj.exe. LINK386 does not understand .o object code. I hope that the
next version of GCC/2 will be a bit more flexible in this regard as I
intend to get support for cross compilation working. I will also be
working to get some more meaningful error messages other than 'code 6'
and stuff. 

BTW, code 6 means that the driver failed when trying to execute a
program. This can be due to wrong path variables, or trying to invoke
ld by using gcc -o foo.exe foo.c instead of gcc -c foo.c (or similar).

>
>BTW: Has anyone noticed that the GCCv2.2.2 present on hobbes is lacking some of
>its header files? These are (again, my memory isn't the best) the ones in the
>gcc-include/machine directory.

ftp-os2.nmsu.edu:/pub/os2/2.0/programming/gcc2-2.2.2/gcchdr.zoo now
contains all neccessary header files. Again, apologies for the oversight.

Trevor Lampre
<mmtl@cc.flinders.edu.au>
Flinders University of South Australia

P.S. The tentative release date for the next version of gcc is early
October. Although I will try and meet this deadline I have been unable
to do any work for the last two weeks due my son being ill and the
death of my father. We will also be moving into out new home soon so
this is a rather tiring time for my family and myself.