[comp.unix.i386] Compiling g++ under SCO

mykel@saleven.oz (Michael Landers) (08/16/90)

Hardware:	Olivetti CP486
OS:		SCO Unix V3.2
g++:		1.37.2	(i386-sysv)
libg++:		1.37.2
gcc:		1.37.1	(i386-sysv and i386-sysv-gas (see below))

I have currently got gcc compiled and running beautifully (it's not using gas
at the moment).  G++ compiles ok, as does libg++, assuming I tell it to use
COFF type stuff.  Unfortunately, when I try to compile something else (it was
groff, actually) I get problems with multiply defined symbols.  These occurs
when there are two static members of different classes with the same name.
I seem to remember I had some other problems with this system, but they
temporarily escape me.

I thought it might be a assembler/linker problem, as groff has definitely
been compiled with g++ (on a Sun, I believe), so I switched to plan B.
I compiled GNU-ld and gas (1.35) and also recompiled gcc to use these with
COFF encapsulation.  At this stage I couldn't convert libc.a to the BSD
type format used by GNU-ld as the conversion program doesn't cope with
multiple data segments in object files.  I tried to modify robotussin (the
conversion program) but then I started getting multiply defined symbols in
libc.a (as opposed to undefined ones from before).

At this stage I have more or less given up, but I would really like to compile
groff.

If anyone has got g++ working _well_ under SCO, could then drop me some mail
and tell me what they did.  Please be as specific as you can.

Thanks in advance.

Mykel.
-- 
 ()                                   \\     Black Wind always follows
|\/|ykel Landers  (mykel@saleven.oz)   \\    Where by dark horse rides,
_||_                                    \\   Fire is in my soul,
Phone: +612 906 3833 Fax: +612 906 2537  \\  Steel is by my side.

rfg@NCD.COM (Ron Guilmette) (08/20/90)

In article <20423@saleven.oz> mykel@saleven.saleven.oz.au (Michael Landers) writes:
>
>...  I tried to modify robotussin (the
>conversion program) but then I started getting multiply defined symbols in
>libc.a (as opposed to undefined ones from before).

I'm not sure if it relates or not, but I also have had trouble with
multiply defined symbols when using the ld++ (aka gcc-ld) provided
with g++ 1.37.??.  I had this trouble on a Sun4/SunOS4.0x system.
I didn't notice this problem when building libg++ however.  I only
saw it when compiling other stuff.
-- 

// Ron Guilmette  -  C++ Entomologist
// Internet: rfg@ncd.com      uucp: ...uunet!lupine!rfg
// Motto:  If it sticks, force it.  If it breaks, it needed replacing anyway.