On 11 July 2017 at 17:03, Justin Forbes <jmforbes@xxxxxxxxxxx> wrote: > On Tue, Jul 11, 2017 at 3:43 PM, Matthew Miller > <mattdm@xxxxxxxxxxxxxxxxx> wrote: >> On Tue, Jul 11, 2017 at 10:26:03PM +0200, Florian Weimer wrote: >>> I ran into this unannounced change: >>> https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels >>> If this is accepted, all x86 hardware on which Fedora can run will >>> support SSE2, and we should reflect that in the i686 build flags. >>> How likely is it that this proposal is accepted? Ideally, we would know >>> this before the mass rebuild so that we can change the compiler flags in >>> redhat-rpm-config. >> >> Currently i686 users are at about 1/6th of x86_64 users, by mirror >> checkins. I don't have an easy way of knowing how many of those i686 >> checkins are old releases -- I'll need to ask Smooge to make a custom >> report -- but I think it's fair to guess that it's significantly tilted >> that way. So, taking a SWAG, I'd say maybe 10% of our users would be >> impacted. That's pretty big, but on the other hand if the cost is >> disproportionate -- and having heard from the kernel people about this >> for several years, I think it might be -- it's probably something we >> should do anyway. >> > > The kernel team quit "supporting" i686 several releases ago, it is > down to community support, which is pretty much nonexistent. Sure, > people file bugs, but rarely do people point to or supply patches for > those bugs. The biggest issue is how much it is ignored by upstream > as well. We have issues where things were never tested on i686, and > then have to be fixed before we can release necessary and relevant > updates which impact everyone else. And discontinuing the i686 build > for F27 would still mean over a year left of supported Fedora on i686 > hardware. > When looking at those check ins, it would also be interesting to note > which of those are running on virt or containers. Containers would > still be possible, and 32bit userspace in virt guests with a 64bit > kernel would still be possible. The kernel header package would still > be built, so all other 32bit i686 packages would continue to build and > work just fine. > Just to head this off. There is no way currently available to determine if a yum/dnf update was done in a virt/container/Workstation/Cloud/Server etc. Attempts to get that data reportable has been stalled for N years. > Justin > _______________________________________________ > devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx > To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx -- Stephen J Smoogen. _______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
