Re: [oe-core][PATCHv2 0/3] Display manager proposal for x11 and wayland

[Randolph Sapp <rs@ti.com>]

On Wed May 7, 2025 at 8:37 AM CDT, Alexander Kanavin wrote:
> On Wed, 7 May 2025 at 01:03, Randolph Sapp via lists.openembedded.org
> <rs=ti.com@lists.openembedded.org> wrote:
>
>> Ah, this is the first I hearing about the gnux32 triplet. Well, that is being
>> filtered out by COMPATIBLE_HOST in the goarch bbclass. Some of those values seem
>> to be invalid now (powerpc64 is supported upstream). That being said, I don't
>> see any mention of the new x32 ABI upstream. Not entirely sure how to address
>> that at the moment.
>
> There are two options:
>
> - set up the x32 build as specified in [1], replicate the fail, tweak
> COMPATIBLE_HOST as needed (in a separate commit) and see if any
> further issues come up.
>
> - correct [1] to no longer build core-image-sato for x32, and replace
> it with core-image-full-cmdline, reducing x32 to non-graphical
> targets. It's been a controversial target for a long time, as no one
> has ever seen actual usage of it, and there's been a stream of
> graphical related crashes and build failures. We keep it around for
> now, but I'd fully support dropping the graphics and leaving only the
> console images.
>
> [1] https://git.yoctoproject.org/yocto-autobuilder-helper/tree/config.json#n867
>
> Alex

Alright, well I did a little work poking around go. It has no understanding of
the x32 ABI currently. Dropping the x32 tests from core-image-sato to
core-image-full-cmdline can work, but that does mask a bit of a regression.

As useful as a display manager will be, I'm not sure I can justify that much of
a regression for an entire ABI. I know the linux kernel has debated about
dropping x32 support a number of times, GCC seems to be the only compiler that
has fully implemented it, and Yocto is one of 3 listed distro solutions that
currently support it. Still feels a little rough. I don't know.

Anyone got any strong opinions here?