| === doko_ is now known as doko | ||
| rsalveti | infinity: slangasek: question for you guys, android gets built on x86, producing 3 'all' and 2 'i386' packages | 03:21 |
|---|---|---|
| infinity | rsalveti: Not anymore. | 03:21 |
| rsalveti | this was working fine when uploading new android packages to utopic: https://launchpad.net/ubuntu/+source/android/20140922-1903-0ubuntu4/+build/6468836 | 03:21 |
| rsalveti | yay, that's probably the answer then | 03:21 |
| rsalveti | noticed my current android package, for vivid, is only producing binaries for 'i386', not all anymore | 03:22 |
| rsalveti | infinity: is amd64 the primary arch now? | 03:22 |
| infinity | rsalveti: amd64 is the arch-indep build arch now. | 03:22 |
| rsalveti | great, that explains then | 03:22 |
| rsalveti | infinity: now the question is, how can I force the i386 build to produce my arch-indep packages now? | 03:23 |
| infinity | You can't. | 03:23 |
| infinity | You need an amd64 build. | 03:23 |
| rsalveti | crap, that might not be that trivial with the android package | 03:23 |
| rsalveti | but well | 03:23 |
| infinity | More trivial than finishin all the arch-affinity work to let you specify where you want to build, I'm afraid. | 03:24 |
| rsalveti | yeah | 03:24 |
| rsalveti | thanks anyway | 03:24 |
| rsalveti | was this announced anywhere? | 03:24 |
| rsalveti | yeah, it was, just saw the email | 03:25 |
| * rsalveti just got back from vacations | 03:25 | |
| === Saviq is now known as Saviq-codedive | ||
| rbasak | cjwatson: if you put on your SRU team hat today, please can I poke you about reviewing juju-core in the Trusty queue? I'd like to get it pushed through soon if we can, so that upstream can QA on Ubuntu's proposed binary before they release upstream. | 12:16 |
| doko | rbasak, he's not available this week | 12:30 |
| rbasak | OK, thanks | 12:33 |
| mlankhorst | can the xorg-server upload be removed? I forgot to build with -v so the old bug won't get linked correctly | 14:28 |
| rsalveti | infinity: ogra_: hey, so we can't yet build the android package on amd64, it'd need a lot of work to get that done, so we need to build it on i386, but I believe we also need the final android package to be all instead of i386 | 14:47 |
| rsalveti | as that's what is currently used when extracting the binaries on cdimage | 14:47 |
| ogra_ | right, the binary needs to be arch:all | 14:47 |
| ogra_ | however we achieve that | 14:48 |
| ogra_ | and the build can only happen on i386 unless we fix a lot of android | 14:48 |
| ogra_ | infinity, would there be a way to make the android package i386 but to cheat with the publisher to copy the binary around to become "all" ? | 14:49 |
| rsalveti | one way would be to have at least one package building for 'amd64', and then building android inside a i386 chroot, but that will be ugly | 14:49 |
| xnox | ogra_: no. | 14:49 |
| rsalveti | as currently I only got packages for 'all' and 'i386', and it never gets built on amd64 | 14:49 |
| ogra_ | right, thats the other option beyond having the publisher do something (which probably is equally ugly or requires re-packing>) | 14:50 |
| xnox | rsalveti: actually, it should be not that hard to convert it to build on arch:all on amd64, you need to flip a few dependencies, but otherwise it should work with a multilib toolchain. | 14:50 |
| xnox | unless, well.... we'll need multiarch enabled on the builders. | 14:50 |
| rsalveti | xnox: right, I'd need to install a bunch of i386 packages at build-time | 14:50 |
| xnox | ogra_: rsalveti: actually, src:android should be changed to build i386 arch binary. | 14:50 |
| xnox | ogra_: rsalveti: and then change cdimage et.al. that with vivid it needs to pull an explicit i386 binary..... | 14:51 |
| xnox | =( | 14:51 |
| rsalveti | right | 14:51 |
| ogra_ | ouch | 14:51 |
| xnox | ogra_: rsalveti: or quickly upgrade to lolipop with comes with amd64 SDK =) | 14:51 |
| ogra_ | that will be as ugly | 14:51 |
| xnox | s/with/which/ | 14:51 |
| rsalveti | can't for a few months | 14:51 |
| rsalveti | at least | 14:51 |
| xnox | and full x86_64 support in the emulator et.al. | 14:51 |
| ogra_ | right "quickly" | 14:51 |
| rsalveti | I need something for this week still :-) | 14:52 |
| xnox | rsalveti: well, i believe utopic are still using correct builder - that is i386 | 14:52 |
| xnox | rsalveti: and so does ubuntu-rtm. | 14:52 |
| rsalveti | xnox: right | 14:52 |
| xnox | rsalveti: it's only vivid that builds arch:all on amd64 | 14:52 |
| rsalveti | xnox: yes, but I need this package on vivid as well | 14:52 |
| xnox | rsalveti: you should be able to branch ubuntu-rtm-vivid with arch:all set to i386, but talk to wgrant about that. | 14:53 |
| rsalveti | maybe building on a utopic ppa and doing binary copies? | 14:53 |
| xnox | rsalveti: actually, yeah talk to wgrant about getting this fixed. | 14:53 |
| xnox | rsalveti: yes, building in a utopic ppa will also work and doing a binary copy into the archive. but do it in a non-virt PPA. | 14:53 |
| rsalveti | xnox: yeah, need one utopic silo for that | 14:53 |
| rsalveti | ogra_: do we still have utopic silos? | 14:54 |
| ogra_ | could be, ask sil2100 | 14:54 |
| xnox | or ev. | 14:54 |
| xnox | =) | 14:54 |
| ogra_ | rsalveti, but why does it need to be a silo ;) | 14:54 |
| ogra_ | any native PPA should do | 14:54 |
| sil2100 | We still have utopic silos, yes, but those need to be SRUed anyway | 14:54 |
| rsalveti | ogra_: well, we need a non-virt PPA | 14:55 |
| ogra_ | like canonical-arm-dev ;) | 14:55 |
| rsalveti | hm, last time I tried copying stuff from that PPA I had issues, just don't remember which ones | 14:55 |
| rsalveti | I had to rebuild them on a proper silo | 14:55 |
| ogra_ | hmm | 14:56 |
| rsalveti | sil2100: what happens if I build/push a src package for utopic on a ppa currently used for vivid? | 14:56 |
| rsalveti | sil2100: I guess we only need to clean it up once done | 14:56 |
| rsalveti | ppa/silo | 14:56 |
| sil2100 | rsalveti: you only want to use that silo for building, right? Not for releasing through the train? | 14:56 |
| rsalveti | sil2100: nops, just building | 14:57 |
| rsalveti | will need to manually release it | 14:57 |
| ogra_ | well, for copying later too | 14:57 |
| sil2100 | rsalveti: then if it's one of your silos, you can upload without any problems - just make sure to clean up afterwards | 14:57 |
| sil2100 | i.e. delete the package once done | 14:57 |
| rsalveti | guess I'll upload to utopic in there, then binary copy to vivid on the same silo and will try using the train | 14:57 |
| sil2100 | A silo in the train can be configured only per one series, but that only counts when you want the train to track the upload and release it | 14:58 |
| rsalveti | yeah, guess that will have to do it | 15:00 |
| === seelaman` is now known as seelaman | ||
| teward | can a release manager approve the trusty nomination on https://bugs.launchpad.net/ubuntu/+source/util-linux/+bug/1389321? (requested by psusi on #ubuntu-bugs to be trusty-targeted). | 16:54 |
| ubot2 | Launchpad bug 1389321 in util-linux (Ubuntu) "partx wraps partition offset and length mod 2^32 (2^23 sectors)" [High,In progress] | 16:54 |
| === Saviq-codedive is now known as Saviq | ||
| infinity | rsalveti: Building in utopic and copying is a workable (very) short-term hack, but the package really needs to be buildable on vivid. | 21:10 |
| wgrant | rsalveti, infinity: I'm in a position now where it's trivial to make 'all i386' resolve to 'build arch-indep on i386 regardless of nai', if we agree that that makes sense. | 23:37 |
| rsalveti | infinity: that will happen when we rebase android with latest release (5.0) | 23:38 |
| infinity | wgrant: That might make sense until we get arch affinity in debian/control. | 23:40 |
| wgrant | It's actually easier to do that than not. | 23:47 |
| wgrant | So I shall do so. | 23:47 |
| wgrant | (ie. when I'm looking for the build to do arch-indep on, I'll prefer nai but pick an arbitrary arch from the set if nai isn't usable) | 23:48 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!