In message <Pine.GSO.3.95.990303120113.28811B-100000@jericho>"Eduardo E. Horvat
h" writes:

[ugliness with __arch64__]

>Also, there are some shell scripts, GNU config stubs, and source files
>that use "alpha" to mean 64-bit.

Linux braindamage....

>So, I like the idea.  We could get rid of the `sparc64' MACHINE_ARCH and
>just use the "bittness" macro to figure out what to build.

Okay, this is a bit more complex. I was including cases like the mips
`n32' which Castor Fu is working towards, where an 32-bit useland
(ints, pointers 32bits) runs on a kernel that's 64-bit clean, allows
64-bit instructions, and the compilers use native 64-bit instructions
for 64-bit integers.

Thats not quite the same as LP32 vs LP64, which is (I think)
what you're getting at.

To do both, we'd need two Makefile variables, one for LP32 vs LP64,
and one for `LP32-bit but with native 64-bit insns for 64-bit ints''.

How about these:
	LPSIZE		either "lp32" or "lp64"
	WORDSIZE	either "64" or "32"

I'm not fussed about the names.