Mail Archives: djgpp/1998/09/28/05:40:41
On Sun, 27 Sep 1998, George Foot wrote:
> > > %{Zwin32: \
> > > %{!Zdll: c:/djgpp/rsx2/rsxntdj/lib/crt0w32.o} \
> > > %{Zdll: c:/djgpp/rsx2/rsxntdj/lib/dll0w32.o} \
> > > %{Zsysv-signals: c:/djgpp/rsx2/rsxntdj/lib/sigsysv%O} \
> > > %{Zbsd-signals: c:/djgpp/rsx2/rsxntdj/lib/sigbsd%O} \
> > > %{Zbin-files: c:/djgpp/rsx2/rsxntdj/lib/binmode%O} \
> >
> > I think it's a bad idea to include absolute path names here. Why
> > isn't it possible to lump these files into DJGPP's lib directory and
> > then have specs call them by their basenames?
>
> I agree that it's bad to specify the full paths, but I also
> don't much like the idea of mixing up all the libraries.
Sorry, I must be missing something: where's the mixup? All of the
names above (crt0w32.o, dll0w32.o etc.) don't happen at all in the
stock DJGPP distribution. So what's the problem to have them all in
the same directory with DJGPP's libraries and object files?
> The problem will still exist though because the
> linker needs to know which version of libc.a to use, and the
> preprocessor needs to know which include files to use.
Will renaming the libraries that have the same names (if there are
such) and using lib/specs to force GCC to use the correct ones solve
this problem?
- Raw text -