Re: Clean up MD_EXEC_PREFIX and MD_STARTFILE_PREFIX definitions

[Rainer Orth <ro at CeBiTec dot Uni-Bielefeld dot DE>]

"Joseph S. Myers" <joseph@codesourcery.com> writes:

> This patch, relative to a tree with my AS_NEEDS_DASH_FOR_PIPED_INPUT
> patch <http://gcc.gnu.org/ml/gcc-patches/2010-12/msg00362.html>
> (pending review) applied, cleans up MD_EXEC_PREFIX and
> MD_STARTFILE_PREFIX definitions in GCC, eliminating those in svr4.h.
>
> The definitions in svr4.h would have been inherited by many
> configurations that did not undefine the macros - but most of those
> were cross-only configurations and the macros are native-only.  As
> well as bare-metal configurations, I believe the following affected
> target OSes are cross-only (based on the absence of code in
> config.guess to generate associated triplets): VxWorks, RTEMS,
> uClinux, Symbian, Netware, TPF.  Thus, the definitions in svr4.h were
> only relevant to QNX, Solaris and IA64 HP-UX; all other configurations
> using svr4.h either undefined those macros, or were cross-only.  This
> patch moves the definitions to more specific headers for those
> targets, removing #undef directives in other headers that are no
> longer needed.

I'm not sure this is the right approach, actually: those locations
(/usr/ccs/bin and /usr/ccs/lib) are the generic System Vr4 directories
and it makes perfect sense to me to have them defined in a generic
svr4.h header file.

On the other hand, many other OSes seem to include svr4.h though they
have no business at all doing so.  It seems to me that it would be
better removing those spurious includes rather than duplicating the SVR4
definitions for those systems they are right for.

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University