Re: PATCH support for FSC machines.

[jean-frederic clere]

Robert Boehne wrote:
> Jean:
>
> It looks like the first thing that the patch does
> is remove a clause that conditionally sets lt_cv_prog_cc_wl to
> '-LD'.  I have some misgivings about doing that without
> knowing why that was put there in the first place.
> If you know when or where '-LD' is appropriate, can you
> set that where it is?  Perhaps this is particular to a
> certain version of a certain compiler, if so it should
> be set based on that.  Perhaps we'll hear from someone
> who knows where it came from...


I think -LD was an error.
I have fowarded the mail to a colleague and he says the following:
+++
Actually, I think that it was a bug all the time. No compiler since the
days of SINIX-1.x ever used -Lanything for purposes other than specifying
the lib search path. (-L/usr/lib -L/usr/ccs/lib ...)
So -LD means: look in the subdirectory ./D/ for any libraries requested
by -lanything
++++

He also says that probably not all the compilers were able to pass a string to 
the ld via cc.

> diff -ru libtool/libtool.m4 libtool-1.4.2.ok/libtool.m4
> --- libtool/libtool.m4  Tue Sep 11 05:18:52 2001
> +++ libtool/libtool.m4  Tue Mar 19 14:25:49 2002
> @@ -952,11 +952,7 @@
>      sysv4 | sysv4.2uw2* | sysv4.3* | sysv5*)
>        lt_cv_prog_cc_pic='-KPIC'
>        lt_cv_prog_cc_static='-Bstatic'
> -      if test "x$host_vendor" = xsni; then
> -       lt_cv_prog_cc_wl='-LD'
> -      else
> -       lt_cv_prog_cc_wl='-Wl,'
> -      fi
> +      lt_cv_prog_cc_wl='-Wl,'
>        ;;
>  
>      uts4*)
>
> Thanks,
>
> Robert