Alexander Graf - 12:53 5.01.16 wrote: > > > > Am 05.01.2016 um 12:07 schrieb Michal Hrusecky <[email protected]>: > > > > Andreas Färber - 9:50 5.01.16 wrote: > >> Hi Michal, > >> > >>> Am 05.01.2016 um 08:52 schrieb Michal Hrusecky: > >>> I was searching but wasn't able to find where lives Leap for armv7l. Or > >>> is it not ready yet? > >> > >> Apparently you didn't search the list archives: > >> http://lists.opensuse.org/opensuse-arm/2015-11/msg00011.html > > > > Yep, I was searching in OBS, didn't expected answer this bad. > > FWIW there is no 32bit Leap anywhere - not on x86, not on ppc and not > on ARM.
Yep, I know and on x86 it kinda makes sense, on ARM, most of the boards out there are still 32bit, so it makes sense not to drop 32bit ARM for openSUSE (I understand reasoning for SLE though). > > > >> A workaround for the Arndale boards has been identified by Bernhard and > >> Dirk during Hackweek, but I'm not sure whether it's deployed yet. > > > > Ok, will try to read up what was the issue. > > > >>> If so, where lives wip so I can help? > >> > >> If you want to help, set up a private OBS instance somewhere and start > >> building Leap and fixing any failures. Chances are that if you use > >> qemu-linux-user on x86, like Factory's armv6hl, you'll be able to build > >> packages but won't be able to build any images beyond the rootfs. > > > > Hmmm, in general not enough resources to effectively fork whole OBS in > > long term, don't have powerful enough computer that I can spare for a > > long time. > > > > So you are saying that currently there is no effort to support armv7l on > > Leap, just somehow building Factory. And if I will try to do it in main > > OBS, it will just cog up and there will be neither Leap nor Tumbleweed > > as we lack build power. And the workers cannot be distributed outside of > > NUE datacenter, so I can't contribute my spare CPU cycles. Am I correct? > > I'm fairly sure if someone else takes care of the "keeping Leap > working" bits, we can solve the build power issues. Would you commit > to be the contact point for it? Next month I'm joining new company at new position, so I'm not sure how hard can I commit to anything. But yes, I can volunteer to receive the mails, not sure about fixing stuff, can do some fixing, but no idea how much time I can dedicate and how much stuff will be broken. It can be probably marketed as experimental best effort to lower the expectations. If there would be 32bit ARM fixes, I guess there is no way to get it in Leap now, but probably it could be possible to get it in next one. So for this release I guess the solution would be forking few packages that does not build, but hopefully not many as SLE is still based on Factory which should work. And even if it is old codebase, it should be something like openSUSE 13.1 and that one had armv7 port... I believe that once started, there will be plenty people helping (I believe I'm not the only one that wants to use Leap on armv7l). So if somebody who will start with it is needed and some point of contact to connect volunteers and respond to requests, I can volunteer to be that person. As mentioned, not sure how much time I can dedicate it in near future, but we'll see how far it will get ;-) I think getting even part of the distro is better than nothing. I tried on private OBS (that I can abuse temporally) to build Ring 0 of Leap using 13.2 and almost everything builds out of the box except Perl (and gcc5 is still building right now, but pretty far already so I have my hopes up). As we have Rings now, I think we can go ring by ring. Having Ring1 might be actually enough for some use-cases - if you add few packages you need... What do you think? -- To unsubscribe, e-mail: [email protected] To contact the owner, e-mail: [email protected]
