Re: [PATCH][Ada] Switch gnatbind from C to Ada generation during bootstrap

[Laurent GUERBY <laurent at guerby dot net>]

On Sun, 2009-06-28 at 09:15 +0200, Ralf Wildenhues wrote:
> * Arnaud Charlet wrote on Sat, Jun 27, 2009 at 03:16:12PM CEST:
> > 
> > Also the use of $(MV) -f looks strange, I've never seen this kind of usage
> > (explicit use of -f) in Makefile before.
> 
> FWIW, automake-generated Makefiles use 'mv -f' in the dependency
> tracking code.  The Autoconf manual documents -f to be portable for mv.

Ada various Makefiles don't currently use a consistent definition for
the default MV:

ada/gcc-interface/Make-lang.in:MV = mv
ada/gcc-interface/Makefile.in:MV = mv -f

Before patch file behaviour is output overwrite (by gnatbind), that's
why I added "-f" to MV in the patch, to be sure to match the before
patch behaviour.

However this "-f" is not critical: on most system I think it's the
default anyway and I believe the difference can only show up only in
case of manually interrupted build between gnatbind and MV and then the
error message and manual fixup will be obvious.

Since my latest tests have shown it's possible to fully build Ada
compiler, gnatlib and gnattools C parts with a C++ compiler, I don't
think avoiding MV is worth having one file requiring a C compiler.

Anyway as I said it will be up to Ada maintainers to decide what path
to follow for gcc-in-cxx: I won't work anymore on this since I just
wanted to see following discussions at the GCC summit wether or not the
switch to C++ compiler would be problematic for Ada, and it's clearly
not the case.

So, Ada maintainers please feel free to pick up whatever bits you feel
are needed and rework other when the time comes.

Sincerely,

Laurent