Re: LIBGCC_SPEC

[Jim Wilson <wilson at specifixinc dot com>]

On Tue, 2004-02-17 at 15:05, Matt Austern wrote:
> As Geoff said, I don't believe the "target independent" part.  The 
> munging encodes specific library names and the way they correspond to 
> specific options.  There are even comments in gcc.c saying that some of 
> this is just guesswork and that it's probably wrong on some targets.

I still haven't seen an explanation of why this is wrong for darwin. 
You need to explain what you want to do which the code does not already
do, or which the code does wrong.  Otherwise, all of this is so abstract
that I can't understand your problem.

The only assumption is about filename endings for libraries, and that
can be fixed if necessary.

> Munging spec strings strikes me as deeply wrong.

An advantage of doing things this way is that we did not have to rewrite
all of the old LIBGCC_SPEC strings to get the shared libgcc support
working.  This also means that we don't need shared libgcc support in
target code, and hence the support works in a target independent
fashion.

Of course this assumes that we can do shared libgcc in a target
independent fashion.  You stated that this is wrong, but you did not
give any details.

I don't think anyone is dead set against any changes to how LIBGCC_SPEC
works, but you need to provide a convincing argument that the current
mechanism is broken before anything will change, and you haven't done
that yet.
-- 
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com