I don't think it makes sense to push this after the mass rebuild, hence I am -1 for the proposal as written.

If this Change were re-targeted to Fedora 38 (and applied between F37 branch point and F38 mass rebuild) I would be +1.

I don't think that incrementally changing the baseline with bugfix updates during the lifetime of Fedora 37 is a good idea. Hence, I agree with @decathorpe and am -1 (with a potential +1 if this targets F28). However, I am happy to hear why this needs to be done in F37 and why it is safe to change the baseline incrementally after GA.

@churchyard , I believe you mean "+1 for F-38"

• there is 100% compatibility of the new hw with the older ones, eg. z13 will run zEC12 code (the current state) just fine, there is no technical reason to have everything rebuilt
• I would not change the baseline after GA, but after the mass-rebuild and before beta (because I was too late with the change)
• when the kernel has been switched, then it won't start on (any) older hw and it will inform the user
• I won't disagree if the consensus will be to move the Change to F-38

@churchyard , I believe you mean "+1 for F-38"

I do and I edited my post the second I sent it. Sorry about that.

I would not change the baseline after GA, but after the mass-rebuild and before beta (because I was too late with the change)

Packages that are not rebuilt between mass rebuild and GA but have updates issued after GA will change after GA. That is my biggest concern.

I would not change the baseline after GA, but after the mass-rebuild and before beta (because I was too late with the change)

Packages that are not rebuilt between mass rebuild and GA but have updates issued after GA will change after GA. That is my biggest concern.

I see now what you mean, yes, it will be a change, but z13 has been a baseline in RHEL-8 (and EPEL-8 and ELN) for years now without issues, so I believe it won't be a problem. Also the number of possibly affected users is very small, I might know them personally ;-)

Question: why not go to the baseline eln/rhel9 has:
march=z14 -mtune=z15 ?

Anyhow, I am fine doing this for f37 or f38 either one.

Question: why not go to the baseline eln/rhel9 has:
march=z14 -mtune=z15 ?

• I would like to be able to use the existing z13 resources,
• qemu claims they support up-to z13 in the TCG mode (https://github.com/qemu/qemu/blob/master/docs/system/target-s390x.rst)

Anyhow, I am fine doing this for f37 or f38 either one.

thx :-)

Also the number of possibly affected users is very small, I might know them personally ;-)

That may be true, but do you know all the affected packagers who would, possibly for the entire lifetime of Fedora 37, need to work around unexpected build failures on s390x? A build failure on s390x fails the entire build, and since dropping architecture support in a stable release update is "frowned upon", it would basically block updates for those packages entirely.

I stand by my opinion that this should happen only before a mass rebuild, to limit the timeframe in which problems can occur to a few weeks (and not 15 months, as would be the case with switching it now).

Indeed, the low amount of Fedora s390x users makes this a low-impact change for our users, but it is still potentially a high-impact change for our packagers.

This topic is on the agenda for tomorrow's FESCo meeting:

https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/6VAM5N54Z4AJQ7XVYFSZZFC6QTG26P7H/

I'm 0 for this Change. I don't think it would be too significant to change at this point, but I agree that it would have been significantly better to have done it in time for the mass rebuild.

This was discussed during today's meeting (meeting log).

• AGREED: REJECTED for F37 (+5, 2, -1)
• AGREED: APPROVED under the condition that the Change proposal is re-targeted at Fedora 38 and classified as a System-Wide Change (+8, 0, -0)

Announced: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/thread/QI7ARJI4OSJYHBZ56VJWAXM5WDBYLEKL/

I will leave this ticket open for now.