End of 32-bit support?

[Matthew Miller]

On Sat, Jan 24, 2015 at 08:37:59AM +0100, Ralf Corsepius wrote:
> >Hatters, or from Red Hatters working in their spare time. (Of course,
> >as RH often does, many of the high-output contributors end up applying
> >for and getting RH jobs, skewing the picture.)
> Well, I am observing quite a few people from major enterprises (RH
> business partners?) who are working on secondary archtectures, but
> I've very rarely (I don't recall any such incident) tripped over
> community folks who are working on them.

Sometimes Red Hat business partners, but that doesn't mean that it's at
Red Hat's direction. Overall, this is one of the few areas where we
have money and paid effort flowing into the project that *isn't* coming
from Red Hat, and I don't think that's a bad thing. These are
"community folks" too, at least if we're doing it right.

> >Additionally, I'm not privy to Red Hat's architecture strategy, but as
> >far as I know, 32 bit ARM — currently our only primary non-x86 arch! — is
> >not of particular corporate interest.
> It's obvious to me the aarch64 is RH's business interest.

But aarch64 and 32-bit arm are _completely_ different architectures.


> >I also think it's a little unfair to frame this as a conflict, overall.
> >It may be the case that Red Hat is less interested in paying people to
> >work on 32-bit x86 (although I don't actually know that to be a fact).
> >But this is just like any other contributor to the community — you
> >can't make people do work they're not interested in.
> Right, but that's not my point:
> My points are:
> - I once more feel pushed/tossed around by RH's interest and
> RH-Fedora-people who obviously don't properly separate RH and
> Community.

I can't argue with feelings, but I also am not really sure what
separation you're looking for here and how it would affect this.

> - Support for i386 falls out as a by-product at almost Zero-costs of
> the existing process.

I don't think that's true at all. It signficantly increases QA load,
and we're struggling a lot with release engineering being able to cope
with Fedora at its current scale. Cutting back here has an clear
benefit (whether or not it's significant enough to outweigh the other
wide isn't settled, of course). More significantly, the Fedora kernel
team tells me that _they_ don't feel like they have the resources to
really honestly support the 32-bit kernel — and the rest all falls out
from that.

> - Making the i386 a secondary arch will cause additional costs and effort.

As does any change, sure.

-- 
Matthew Miller
<mattdm at fedoraproject.org>
Fedora Project Leader