[01:34] * infinity grumbles about people packaging new libraries without multi-arching them. [02:06] infinity: I know little apart from a 1 hour session of multi-arching an rpm (Red Hat course), but I do agree with you 100%. === jbicha is now known as Guest52408 [04:03] phillw: RedHat doesn't do multiarch, I suspect we're talking about two different things. [04:05] infinity: okies, I was thinking about when I did the building of an .rpm (similar to a .deb) [04:06] phillw: Yes, I know what RPM is. ;) [04:06] lol [04:06] phillw: But "mutliarch" refers to coinstallable packages from different architectures based on unique paths. RedHat doesn't do that. [04:07] So, I assume you were perhaps referring to "making it build on more than one arch", which many RH/Fedora people never bother to think about (since they have a "primary arch / everyone else we don't care about" split), but it sort of the status quo for Debian and Ubuntu. [04:07] reminder to self... leave the devil of the details to the devils :) I was just trying to get a better understanding of how these things are done. [04:18] infinity: in the brief hour I had on it, we could chose to build i386, X64 or for both. It's not a big part of the RHCE course, we just need to be aware of the difference of building for one arch, and the use of multiarch as per .prm [04:18] *rpm* [11:00] cjwatson: Hey, does sharand still belong in multiverse ? [11:01] Daviey: well, it's in Debian non-free; unless we have some reason to care I'd rather they moved it first [11:03] * ogra_ glares at nusakans processlist [11:03] cjwatson: ah, i see.. I just saw references of the licence change.. [11:03] cdimage 6104 6101 0 08:31 ? Ss 0:00 /bin/sh -c DIST=precise for-project ubuntu cron.daily; PROPOSED=1 buildlive ubuntu daily-live precise && DIST=precise for-project ubuntu cron.daily-live [11:03] ... [11:03] dimage 5494 5480 0 08:42 ? S 0:00 ssh -n -o StrictHostKeyChecking no -o BatchMode yes buildd@celbalrai.buildd /home/buildd/bin/BuildLiveCD -l -A armhf -s mx5 -p -d precise ubuntu [11:03] ... [11:03] Yeah, AFAICS that made it distributable but not free [11:03] why the heck is it building precise live images :/ [11:04] in general? because we will care about a point release in the near future [11:04] well, i only see mx5 atm [11:04] for mx5? that is a little confusing [11:04] massively, yes [11:05] see the line above [11:06] ogra@nusakan:~/branches/cdimage-private$ CDIMAGE_ROOT=/srv/cdimage.ubuntu.com/ ALL_DISTS=precise bin/default-arches ubuntu daily-live precise [11:06] amd64 amd64+mac i386 powerpc [11:07] seems ok though [11:08] Yeah, I'm just digging now [11:08] really bad is that it runs 1h after the quantal live build starts ... so it will block the whole until all arm builds are done [11:10] I'm fixing it now [11:10] what is it ? [11:10] * ogra_ is bothered that he couldnt find a cause [11:10] DIST isn't passed through to default-arches [11:10] ouch [11:11] Because the buildlive interface is anomalous and takes it on the command line [11:11] I think I'm just going to regularise the buildlive interface [11:11] yeah [11:11] Though I guess I could just set DIST earlier [11:11] actually yeah, that would work [11:11] shouldnt BuildLiveCD set it ? [11:11] or is that to late [11:11] er that's the next layer down and doesn't help [11:12] ah, k [11:12] FWIW the way I investigated this was to make a copy of buildlive and add 'echo "$ARCHES"; exit 0' just after sourcing config [11:12] then I could sh -x it [11:12] ah ! [11:12] * ogra_ *again* didnt think of -x [11:12] damned [11:14] yeah, well, it only helps if you have a version that won't go on and build stuff :) hence the temp copy [11:15] fixed now; although I do think we need to rebalance the builders again - what happened to different subarches being on different livefs builders? [11:15] I guess that died when we shifted off the beagle kennels [11:16] there is nothing to rebalance with [11:17] we only have celbalrai atm [11:17] i was told elmo works on getting another mandabox up though [11:17] and that we should get balanced builders from that setup then [11:17] not sure where that stands [11:18] fair enough [11:30] cjwatson: Do you have some capacity today to look at squashfs for server? [11:39] Daviey: maybe; how far have you guys got so far without me? [11:42] cjwatson: It feels barely off the starting block.. i was doing debug builds on nectarine.. but when tested, it didn't seem to provide useful info, or find the squashfs.. but did have it on the image. [11:42] cjwatson: Can we cover again what changes you thought were required. [11:43] (it's only really me been working on it) [11:47] In that case I might as well JFDI I guess [11:49] although *nectarine*? why that? [11:49] it shouldn't be involved [11:50] DYM nusakan? [11:50] I see http://people.canonical.com/~ubuntu-archive/livefs-build-logs/quantal/ubuntu-server/latest/livecd-20120704-amd64.out [11:52] we'll need to substitute live-installer for bootstrap-base, which is a bit tricky to do without committing all image builds to it; I might do some kind of temporary hack in cdimage for that [11:54] huh, why did that end up in NEW - it was already in -proposed [11:55] oh, blast, I didn't realise that was part of a transition of some kind [11:55] * cjwatson peers [11:55] I hate this crappy half-use of proposed [11:58] OK, this is crazy, there's no way that actually dropping python-software-properties can work in the short term given some of its rdepends (cloud-init, packagekit-backend-apt, ubuntu-orchestra-client-juju) - I'll revert the packaging to produce both [11:59] cjwatson, cyphermox said those just need their depends adapted [11:59] cjwatson, he said he would chase people down today, he couldn't yesterday (4th of july) [12:00] I'm going to revert it anyway for now to make things smoother [12:00] not convinced of the correctness of the end of this port either given that some files have pretty fatal pyflakes errors [12:01] in any case there's no reason to do a rough transition when a smooth one is possible [12:26] < cjwatson> DYM nusakan? [12:26] yes, sorry. [14:00] Daviey: do you have a decision on what you want to be in the squashfs? I assume that server being what it is you probably don't want to ship everything on the CD in the squashfs and would like some other things to be shipped as debs? [14:00] is ubiquity-dm known to have any issues atm ? on omap images i end up in a live session (and get an apport popup for ubiquity-dm) [14:02] Daviey: and do you want broadly the same set of packages to be on the converted image, one way or another? [14:05] ogra_: not sure it's known, but I see that here too (omap4) [14:06] k [14:06] * ogra_ doubts he can run an install on a beagle if the live session already ate all my ram [14:07] ogra_: installing will fail anyway because of ubiquity exploding when installing langpacks [14:08] ah, crap, k [14:08] * ogra_ pulls a netboot image then [14:09] (and dont tell rick :P ) [14:09] cjwatson: Yes.. just server in the squashfs, server-ship still as debs.. is that possible? [14:12] should be [14:12] first attempt produced a tiny image so I obviously got things wrong [14:17] geez, on a begle the netboot installer on fb really hurts my eyes ... somehow the beagle interprets violet backgrounds as bright neon pink [14:18] ogra_,infinity: do you think I'm OK to add the server seed onto the ubuntu-server preinstalled images, as a side-effect of what Daviey's asking for? i.e. http://paste.ubuntu.com/1076482/ [14:18] cjwatson: I managed to create a 400MB image IIRC [14:19] cjwatson, well, i think we would like to drop preinstalled for server once Daviey is done ... and switch arm to the new installation too [14:20] so it shouldnt really matter [15:07] cjwatson: I recall advice from you not to worry overmuch about packages that weren't useful in Ubuntu, but were sync'ed from Debian because it wasn't worth the trouble to maintain the blacklist. Now that the blacklist is more self service (for me anyway, I don't recall who can access it), is there a problem with being more expansive about stuff we remove (specifically in this case chkconfig, which doesn't work with upstart and won't)? [15:09] who> ~ubuntu-archive. That's probably OK - just try to keep blacklisting to stuff that will never be useful, rather than that is a bit busted at the moment [15:18] OK. Thanks. [15:19] In particular there's no need to blacklist for fakesyncs or already-removed-from-Debian or such - though you probably aren't in the habit of doing that since you've only used the new sync/removal tools [15:19] Right. No, I haven't. === ralsina is now known as Grumpy === Grumpy is now known as Grumpy_ === Grumpy_ is now known as ralsina [15:55] cjwatson: talking of fakesyncs, do you do them when auto-sync reports that they need to be done? [15:55] or do we have a report listing teh packages out of sync because of tarball mismatches [15:57] tumbleweed: I just do them [15:57] It dodesn't list them separately but I generally notice [15:57] as long as tehy aren't going unnoticed, that's what matters [15:57] Let me do a final auto-sync for quantal just to check [16:15] cjwatson: If preinstalled sticks around, no, that change would be "wrong", IMO. [16:16] cjwatson: At least, I assume that "server" is being added to the squash build as the removable "live" part? [16:16] cjwatson: If not, then Daviey needs a talking to, cause not every server install in the world should have apache and mysql by default. [16:17] in the preinstalled case i really dont care [16:17] assuming we actually drop them anyway [16:17] ogra_: If it sticks around, I do. :P [16:17] ogra_: If the server live experiment works out, then yeah, we'll drop preinstalled and switch to that. [16:18] (At which point, I won't care) [16:18] well, right, but the initial idea of offering people a minimal dev system through the arm server images is moot anyway [16:18] ? [16:18] the download is to big, most people use netinst anyway [16:18] and i dont really know any actual "server" users [16:18] Oh. I thought you meant it wasn't "minimal" enough, which it is. [16:18] But yeah, the image is large. [16:19] right, and because it is large we can as well preinstall some server apps [16:19] i wouldnt mind [16:19] I would. [16:19] And it violates a longstanding Ubuntu policy. [16:19] to have useful bits preinstalled ? [16:20] honestly i think though that we should probably drop server completely and resort to netinst [16:21] (for arm that is) [16:21] infinity: server isn't what you think it is [16:21] infinity: are you thinking of server-ship? [16:22] the server seed is what we install by default on servers [16:22] doesnt server define a set of apps that get installed ? [16:22] go and look :) [16:23] cjwatson: Oh, I misplaced my ^ [16:23] Note, selecting 'serverstats' for regex '^server' [16:23] ^-- That doesn't end well. [16:23] hah [16:24] cjwatson: In that case, yeah, it should match on preinstalled, for all that it doesn't matter much either way. [16:25] oh, it does install additional apps ... ! [16:25] 10 of them :P [16:25] Yeah. [16:25] right, thats fine i guess [16:25] Hasn't kirkland been gone long enough that we can drop byobu from the server seed now? ;) [16:25] though i would still like to drop arm server images :) [16:26] ogra_: I'd like to drop pretty much all ARM images other than our "demo" desktop image for the blessed platform of the week (which is still omap4 right now). [16:26] ogra_: But we've had this discussion. [16:26] indeed [16:27] And given that netboot is the server experience we're giving people with real server hardware, it seems silly to offer something different for people without. [16:27] yeap [16:27] We really should discuss this with a broader group of people before I just go mad and turn all the images off. [16:27] we should, i'll prepare a mail for tomorrow to ubuntu-devel [16:27] (and will ping the eilt people about it) [16:28] lets see what the feedback is [16:28] Pass it by me before you send it? [16:28] sure, np [16:28] I'll pass it through my "nutty German" filter. ;) [16:28] *duck* [16:28] haha [18:40] cjwatson: ^ there you go [19:57] Does this mean ABI compatibility is hard: http://www.technolog.msnbc.msn.com/technology/technolog/why-some-your-recently-updated-iphone-ipad-apps-are-crashing-864534 [20:03] cjwatson: What is the current status of the squashfs.. is there anything i can help with? [20:04] ScottK: Is that why we opted out of it? :) [20:10] cjwatson: Hrm. I just accidentally removed a package without a comment. I'd perhaps argue that remove-package shouldn't allow that. [20:10] cjwatson: Thoughts? [20:32] infinity: I thought remove-package /didn't/ allow that...? [20:40] slangasek: The LP version might not have, the API one sure just did. [20:40] cute [22:28] Daviey: I ran out of time today but will continue tomorrow; I've done some necessary seed mangling, need to get the correct set of seeds onto the image [22:28] stgraber: ust> thanks [22:29] infinity: feel free to fix the client to prevent that - or submit an LP branch to prevent it if you feel enthusiastic [22:29] though pretty sure the former will be quicker :) [22:32] cjwatson: thanks muchly.. if i can help, please do ask. [22:32] cjwatson: Yeah, fixing the client seems reasonable for now. I'm not sure the API should flat-out block it anyway (though maybe it should). [22:33] Daviey: it shouldn't be *desperately* hard from here [22:34] Daviey: but I've blocked out a couple of hours for it on my calendar anyway to try to make sure I don't forget [22:35] cjwatson: really appreciate your effort on it. [22:36] cjwatson: I'd like to work out where i was being daft tho. [22:37] main thing was probably making sure that the image has live-installer.udeb on it rather than bootstrap-base.udeb [22:37] at a guess [22:37] which is not entirely obvious, but I have a hack in place to let me produce images that way [22:38] ah! [22:38] that probably explains why it simply wedged with no useful debug [22:38] right [22:38] hopefully tomorrow it'll be debug-into-existence territory [22:38] heh [22:38] but I must sleep now [22:38] thanks cjwatson ! [22:40] three LP branches landing in parallel now that should give us the final bits of API queue [22:40] then I get to remove 2000 or so lines of code once that's deployed :)