At 03:29 PM 10/13/2001 -0400, James Mastros wrote:
>On Sat, 13 Oct 2001, Simon Cozens wrote:
> > A nice idea, but I don't think it's going to happen; we'll really need
> > PMCs and string registers for anything sensible (which is why I'm
> > being really quiet this weekend and trying to get PMCs implemented)
>I wasn't thinking so much of gas being the normal assembler, though (GNU)
>ld would make a quite nice linker, I should think, since it already has a
>large part of the hard work done.
I'm not sure we can use GNU's ld, though it would be nice. Our requirements
are a bit different than most other languages. (Though I suppose if they
can manage to link C++, we ought to be a cinch... :) Plus our linker's part
of the runtime--we really do need to link when we start, not at some
external point.
>I was more thinking of it as a first step to letting gcc tarket the PVM.
! I was thinking the other way around. Now you've made my head hurt. :)
> > and
> > we'll also have a variable opcode table, which binutils doesn't like.
>Hm. When we figure out the ABI for calls between routines of different
>opcode tables, we'll have to teach it to gcc.
>
>I can see this as a big, big win. I can also see this as being bigger
>then I can handle, but a damm nice project.
If someone can, that'd be cool, because we could use the same work on other
platforms that aren't using GNU's binutils, since everyone pretty much
needs to do the same sorts of things.
Dan
--------------------------------------"it's like this"-------------------
Dan Sugalski even samurai
[EMAIL PROTECTED] have teddy bears and even
teddy bears get drunk
