Re: [maint] move self to can-commit powerpc-linux

[Mark Kettenis <mark dot kettenis at xs4all dot nl>]

   Date: Mon, 30 May 2005 16:03:54 -0400
   From: Andrew Cagney <cagney@gnu.org>

   Keeping with the general trend, I might as well have myself listed as 
   "can commit" for PPC GNU/Linux.  Once wart though, that list was burried 
   under "Target Instruction Set Architectures" so I also gave it its own 
   heading (PPC GNU/Linux isn't an ISA, its more of an ABI :-).

What trend do you mean?  I can't say that I find this less confusing
than what was previously there, except perhaps that you're
distantiating yourself from the PPC GNU/Linux native code.

Anyway, I've always considered the *-tdep.c files as implementing an
ABI.  Part of that ABI is the ISA.  Several ABI's might share the same
ISA, and therefore share quite a few bits of code.  But there is no
clear boundary between ISA and ABI.  Many prologue scanners for
example use ABI-specific knowledge.  So I really think the distinction
between ISA and ABI in the maintainers file is artificial and doesn't
really help us.

I think we really have three kinds of non-generic code in our tree:

1) target-specific code
2) host-specific code
3) native-specific code (target==host)

I think it makes sense for us to list maintainers for these three
categories using triplets.  The maintainers that you're associating
with ISA's right now are probably best characterized by a catch-all
for target-specific code for a certain CPU type, i.e. powerpc-*-*.  I
myself consider myself responsible for

target-specific code
--------------------

i?86-*-*
m88k-*-*
sparc*-*-*
vax-*-*
x86_64-*-*
*-*-openbsd*
*-*-freebsd*

host-specific code
------------------

*-*-openbsd*
*-*-freebsd*
i?86-*-linux*
x86_64-*-linux*

native-specific code
--------------------

*-*-openbsd*
*-*-freebsd*
i?86-*-linux*
x86_64-*-linux*

The only problem here is that the word "target" is a bit ambiguous
within GDB since we also use the word for

Mark