Re: Another bogus change

[kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)]

    Here `type' is used, but not declared, in the case that
    PROMOTE_FUNCTION_RETURN is not defined.  So, with this change we don't
    even build the first-stage cc1 on an i686-pc-linux-gnu box.

Oops.  The problem here was that emit_group_load has two extra args in 2.95
than 2.8.1.

    I know you were just merging stuff from the GCC2 branch, but it still
    might have been a good idea to check this stuff in in smaller,
    better-tested chunks.

I thought about that, but it wasn't clear that there was an obvious way
of breaking them up. The changes themselves have been run for a while,
so the major source of bugs would be in cases like this one where the
change needed conversion.  I got a lot of those out with my bootstrap,
but that does indeed leave other machines.

I knew there was a good chance of breaking something, but I think it
unavoidable given the number of changes and the number that needed 
retrofitting.