Re: Patch to decl2.c

[Mark Mitchell <mark at codesourcery dot com>]

> I've been trying for the last two days to un-nest these
> function compilations, so that instantiate_decl just queues
> the new function and we compile it on the next round of
> toplevel queue processing.  

That should, in theory, be OK.  We want to instantiate the function,
including its body, in the /*defer_ok=*/0 case, but actually generating
code in the .s file can wait until later.

> I havn't gotten that to work yet.
> Seemingly unrelated changes cause the parser to go haywire,
> and tracking the problems down one by one is tedious.

Hmph.  If you have a specific question about the parser, I can certainly
try to answer it.

> In the meantime, I think the following may work.  I'm just 
> now starting a gcac bootstrap, so I don't expect results 
> until after I've quit for the day.

Yes, that's looks like what I was suggesting in my follow-up email; it
*should* work, and it has none of the characteristics that I was
objecting to about Geoff's patch.

Thanks,

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com