Re: -mmacosx-version-min and libgcc

[Ian Lance Taylor <ian at airs dot com>]

Geoffrey Keating <geoffk@apple.com> writes:

> I'm not gating anything on anything.  I was just pointing out that
> your statement
> 
> > It doesn't have to have precisely the same value as
> > it does at top-level; it just has to work well enough for developers,
> > not for users.
> 
> is not correct; it has to work for users, since it's what gets used
> in the bootstrap.  So it's not as trivial as all that.

I was talking about something different from what I think you are
talking about, though.  I agree that you have found a bug.  I suspect
that the proper fix for that bug is either top-level bootstrap or not
clearing MAKEOVERRIDES, or both.

At the time MAKEOVERRIDES was added to gcc/Makefile.in (1994) we
actually did pass several variables down in the recursive make (not
AR_FOR_TARGET, though).  We had to do that because for make programs
that was the only way to get the variables defined correctly, and
since we were explicitly passing the variables we didn't need
MAKEOVERRIDES (and it caused problems on systems with limited execve
environment space, like SCO Unix).  I think that now that we use GNU
make and no longer use pass many variables down in the command line,
it is safe and correct to stop defining MAKEOVERRIDES.

In any case, I'm glad you're testing the patch.  Thanks.

Ian