Re: lang/gcc* and MULTILIB: arm64 (aarch64) is never set up, even for main

On Feb 9, 2024, at 13:35, Gerald Pfeifer <gerald@pfeifer.com> wrote:

> On Fri, 9 Feb 2024, Lorenzo Salvadore wrote:
>>> [lang/gcc*] never adds MULTILIB for aarch64 (arm64).
>>> 
>>> But these days arm64 (aarch64) has lib32 support:
>>> 14.0-RELEASE has it and main [so: 15] has it.
> 
> Does anyone use this, though? (It's not like FreeBSD provides proper 
> bi-arch support with ports and everything.)

How is amd64/i386 support status any different for such
considerations? Are you suggesting removing those? (You
may well be.)

So far, I do not see much of a justification for non-uniform
treatment of amd64/i386 vs. aarch64/armv7 for this subject
area for now. (For these, both involve tier 1 as well.) armv7
is projected to still be tier 2 for 15.x . i386 is projected
to be unsupported. For non-uniform treatment, that could lead
to swapping the status.

(32-bit powerpc and powerpcspe are projecgted to be unsupported
for 15.x .)

>> I think this is not deliberate, but I have never touched this block
>> of code since I took the GCC ports maintainership yet. We can
>> probably add MULTILIB for aarch64 too and I will check into it
>> as soon as possible.
> 
> When that code was written, aarch64 did not feature lib32 support. Is 
> this, and support by lang/gcc*, "academic", or something users really 
> want and are likely to use?

lib32 was deliberately added. How much do the justifications for
that imply something for this subject area? (lib32 status is not
a platform if itself. chroot/jail support may well not be a
platform of itself: not a platform unless it boots?)

clang/clang++ only may well be viable. But I'm still looking for
a reason for non-uniform handling amd64/i386 vs. aarch64/armv7 .
(That does leave 2 directions: removal vs. adding.)



===
Mark Millard
marklmi at yahoo.com