**** BEGIN LOGGING AT Thu Oct 18 03:00:02 2012 Oct 18 10:17:23 heyho! Oct 18 10:17:45 jama, mrmoku: may one of you please grant me rw acces to phonefsod.git? Thanks! Oct 18 13:46:55 do we want shr branches of oe-core/meta-oe to still track master or do we switch to danny release? Oct 18 16:10:13 mrmoku: what been that onleliner 'trick' to upgrade when we hadn't enough RAM to do it in one big chunk? Do you remember? Oct 18 16:11:13 something like >> for p in $( dpkg -l ); do dpkg -i $p; done < ooh opkg Oct 18 16:12:07 anybody else ? Oct 18 16:24:30 JaMa, hey! could you please grant me access to phonefsod.git ? Oct 18 16:24:47 (rw access i mean) Oct 18 16:35:38 DocScrutinizer05: opkg list-installed | xargs -n 1 opkg install Oct 18 16:35:40 DocScrutinizer05: should work Oct 18 16:36:21 On Debian that would mark every single package as "manually installed" though. I do not know about how that'd be handled by OE. Oct 18 16:36:31 DocScrutinizer05: long time no see, how are you doing? Oct 18 16:36:54 headdesking about funny tasks in job Oct 18 16:37:00 hi PaulFertser :-D Oct 18 16:37:05 thanks Oct 18 17:39:53 slyon_: sure Oct 18 17:41:11 slyon_: done Oct 18 17:56:33 JaMa, thank you! Oct 18 18:01:35 do we want shr branches of oe-core/meta-oe to still track master or do we switch to danny release? Oct 18 18:01:46 slyon_: GNUtoo-x60: mrmoku: ^ Oct 18 18:02:03 hmmmm Oct 18 18:02:37 maybe we could try to track master Oct 18 18:02:41 not to diff too much Oct 18 18:02:54 danny is the new denzil-like branch? Oct 18 18:03:41 yes Oct 18 18:04:06 ok Oct 18 18:07:24 hmm.. my OE knowledge is pretty out of date, but basically I'd vote for "danny" because a release branch benefits stability. I don't know how stable SHR is these day (as I'm using/working on SHR on Debian) Oct 18 18:08:33 stability vs diffing..... Oct 18 18:08:40 I'm not sure we have the manpower to diff Oct 18 18:08:45 GNUtoo-x60: not diffing Oct 18 18:08:55 GNUtoo-x60: we would stay close to danny branch Oct 18 18:09:01 how? Oct 18 18:09:10 just stop upgrading oe-core and meta-oe Oct 18 18:09:20 build against danny Oct 18 18:09:23 for instance if there is a bug we merge it in master and then put it back in danny Oct 18 18:09:37 do we have the manpower for that Oct 18 18:09:41 yes and send pull request to be back-ported to danny upstream Oct 18 18:09:41 well.. if we'd provide stable/useable images we will probably get more devs soon (there are ~200 gta04 beeing delivered these days) Oct 18 18:09:42 or will we just fix locally? Oct 18 18:09:52 yes I read Oct 18 18:10:17 the point is, that OEBasicHash will became mandatory together with PR server in 1.4 timeframe Oct 18 18:10:17 btw I'm porting some games to meta-games Oct 18 18:10:29 hmmm Oct 18 18:10:41 JaMa, what is you opinion? Oct 18 18:10:51 I don't want to switch to OEBasicHash with current infrastructure while tracking master (lots of rebuilds) Oct 18 18:11:06 hmmm Oct 18 18:11:15 slyon_: do danny branch in meta-smartphone and build official images only from danny branches Oct 18 18:11:45 hmmmm Oct 18 18:11:58 how will we test the fixes against master? Oct 18 18:12:00 because with every oe-core/shr upgrade I have to fix a lot of stuff in meta-oe Oct 18 18:12:26 and I don't know if I'll have time for that Oct 18 18:12:39 ok Oct 18 18:13:03 GNUtoo-x60: build against master branches yourself to test it before sending it upstream Oct 18 18:13:07 what would work in the long term but still require less manpower? Oct 18 18:13:11 yes. sounds good to me, to stick with danny Oct 18 18:13:28 because we clearly have a manpower issue Oct 18 18:13:40 danny branch is less manpower Oct 18 18:14:09 and more manpower to meta-smartphone repo Oct 18 18:14:19 GNUtoo-x60, we can just switch to a new release once it's ready. we have to do some fixes/ports for this new release once every release cycle Oct 18 18:14:40 where we honestly don't have a lot changes lately because everybody is busy with something else Oct 18 18:15:09 fwiw angstrom is uzing denzil since it was released Oct 18 18:15:12 I'm comming back tough Oct 18 18:17:10 will danny really bring us more stability? Oct 18 18:17:38 what cannot be backported to danny? Oct 18 18:18:04 e.g. big upgrades Oct 18 18:18:13 but I don't think we need those Oct 18 18:18:25 we need to fix bugs we have in our code Oct 18 18:18:41 what about efl? Oct 18 18:18:47 it's always broken.... Oct 18 18:18:49 and for that we don't need distraction from upper layers changing (because I'm trying to keep shr branch close to master) Oct 18 18:19:16 well efl has releases now. i guess the bug fix releases will still make it into the danny branch Oct 18 18:19:35 yes if I prepare them then I'll merge them to danny Oct 18 18:19:41 ok Oct 18 18:19:52 then we could have stable efl without issues Oct 18 18:19:55 so bugs will be fixed but no apis will be broken Oct 18 18:19:55 like fixed icons Oct 18 18:19:59 fixed sound issues Oct 18 18:20:05 I've also an issue with ogg vorbis Oct 18 18:20:26 I'm not saying we will never use master branches (or shr branches again) Oct 18 18:20:38 I'm saying that I would switch official buildserver to danny Oct 18 18:20:45 ah ok Oct 18 18:20:48 then that's ok Oct 18 18:20:50 it's better Oct 18 18:21:08 but do we really have the manpower to track both? Oct 18 18:21:20 and without shr branches as official source for shr builds I don't need to maintain them Oct 18 18:21:50 GNUtoo-x60: there are only a few changes to release branches Oct 18 18:21:53 ah ok I didn't understood that way Oct 18 18:21:59 I tought that we would fork danny Oct 18 18:22:02 so it's more like stick to release + few fixes then track Oct 18 18:22:18 and do like now only that instead of tracking master we would track danny Oct 18 18:22:44 something like shr-danny could be usefull Oct 18 18:22:50 yes Oct 18 18:22:53 that's what I tough Oct 18 18:22:55 but instead we can just lock revs in layers.txt Oct 18 18:23:04 and bump them only when they work together again Oct 18 18:23:19 shr/danny gives us more flexibility Oct 18 18:23:33 yes but forces me to maintain them again Oct 18 18:23:40 yes Oct 18 18:23:56 basically you'll have to do the same work than now Oct 18 18:24:00 but against danny Oct 18 18:24:08 which means testing against master Oct 18 18:24:10 and backporting Oct 18 18:24:14 which is more work Oct 18 18:25:19 yes but a lot of changes I had to do were because oe-core was always moving Oct 18 18:26:01 ah ok Oct 18 18:26:31 but with danny as release branch I would expect e.g. you when you fix vorbis to test it in master and send it upstream yourself Oct 18 18:26:41 ah ouch Oct 18 18:26:53 and then you would backport? Oct 18 18:27:05 and if you confirm that it works in master then we can decide send it in pull-request for danny branch Oct 18 18:27:20 and when it's applied in danny branch we'll bump rev in layers.conf Oct 18 18:28:35 for example last 4 staging feeds had changes mostly from oe-core/meta-oe not really related to SHR images (at least from user perspective) Oct 18 18:28:57 but I had to change more stuff to keep them building Oct 18 18:29:05 ok Oct 18 18:29:18 roughly less work for you and more for me Oct 18 18:29:59 note that I've some issues recently Oct 18 18:30:40 not exacly that, but less work caused by changes we're not really interested in Oct 18 18:30:54 yes Oct 18 18:31:14 anyway it's not that I have a lot of choice anyway Oct 18 18:31:37 I'm using different set of branches then shr from the beginning anyway (jansa/test branches) Oct 18 18:33:10 GNUtoo-x60: we both don't have a lot of choice when OEBasicHash became mandatory.. Oct 18 18:33:27 ok Oct 18 18:33:59 I want be able to even build test shr branches every day like I did now Oct 18 18:34:13 s/want/wont/ Oct 18 18:34:15 JaMa meant: I wont be able to even build test shr branches every day like I did now Oct 18 18:35:01 are there some docs about OEBasicHash somewhere? Oct 18 18:35:21 not really Oct 18 18:35:23 probably on the mailing list Oct 18 18:35:46 but I've explained it here few times already if you're willing to search archive Oct 18 18:36:09 basically it's like an automatic PR bump Oct 18 18:36:38 with all its consequences Oct 18 18:36:54 ok. I'll search the logs for it Oct 18 18:36:59 I'm testing it on nschle's jenkins, and typical upgrade takes 1-2 days to rebuild Oct 18 18:37:11 our buildhost is 5 times slower Oct 18 18:37:17 ouch!!!! Oct 18 18:37:26 puhh Oct 18 18:37:49 so for continuos builds we could upgrade layers only every 10th day or so :) Oct 18 18:38:07 what if the builder is fast but its connection is slow? Oct 18 18:38:33 our builder or some hypotetical builder? Oct 18 18:39:13 out builder has good connection but is really slow Oct 18 18:39:48 it's using swap on shared virtual machine for builds ffs Oct 18 18:40:07 my builder Oct 18 18:40:44 you mean for those tests in master? Oct 18 18:40:45 the problem would be to keep it online 24/24 Oct 18 18:41:09 JaMa, would a network of some jenkins instances be of benefit? We've got runnting this server: http://pss.dyndns.org/ in a frieds basement (50MBit/s connection) Oct 18 18:41:23 I've a core i7 laptop with 6GB of ram, 2hdds7200 raid0+luks Oct 18 18:41:32 *7200RPM Oct 18 18:42:26 slyon_: GNUtoo-x60: those builders are good for your own builds (e.g. against master) Oct 18 18:42:43 yes. thats what i'm using it for at the moment Oct 18 18:42:55 but on official master with whole task-feed, we build a lot of stuff which maybe nobody is using Oct 18 18:43:22 ok Oct 18 18:43:31 so I would say, keep them for your builds as we don't need to build every commit in oe-core on official mirror Oct 18 18:43:38 ok Oct 18 18:44:06 ok Oct 18 18:44:12 I've also a work laptop but I don't think it's allowed to use it for personal stuff Oct 18 18:44:18 else the rest is very slow Oct 18 18:44:35 if you're interested I can still push my branches (like I did for shr branches) Oct 18 18:45:31 but probably less often Oct 18 18:45:36 and maybe with less testing Oct 18 18:47:58 let's do it your way Oct 18 18:48:16 we try and we see? Oct 18 18:48:35 yes Oct 18 18:49:01 I needed the decision now, because oe-core/master now has some commits which wont be in danny Oct 18 18:49:10 yes :) Oct 18 18:49:20 so I wont push my current shr branch with it Oct 18 18:50:51 JaMa, so changes to meta-smartphone should now be pushed to a danny/danny-shr branch, right? Oct 18 18:50:55 let's do it your way means that: Oct 18 18:51:07 the think where we push on master, and backport then Oct 18 18:52:31 hi nschle86! Oct 18 18:52:45 hi Oct 18 18:53:06 nschle86, if you're interested: there's a new stammtisch beeing planned on the stammtisch ML (<- mrmoku as well) Oct 18 18:53:31 will probably on oct. 23rd Oct 18 18:53:38 when and where is it ? my wife has "abendschule sometimes" Oct 18 18:53:56 where isn't decided yet Oct 18 18:54:07 slyon_: right now you can still push to master Oct 18 18:54:16 slyon_: danny branch is for now only in oe-core Oct 18 18:54:48 JaMa, ok Oct 18 18:54:49 slyon_: push to meta-smartphone/shr branch, sorry Oct 18 18:54:55 nschle86, discussion is here: http://lists.goldelico.com/pipermail/stammtisch/2012-October/000012.html Oct 18 18:55:12 Hehe, "abend" == "abnormal end" (like an exit(1)0 Oct 18 18:55:25 slyon_, you mean like in december? Oct 18 18:55:31 we're the 18 oct Oct 18 18:55:32 Sorry, "(like an exit(1))" — it's bad when you typo syntax like that. Oct 18 18:55:35 I forgott about that Oct 18 18:55:56 GNUtoo-x60, no october. it's the munich openmoko user group meeting Oct 18 18:56:03 ah ok Oct 18 18:56:55 there are quite a buch of us here :) Oct 18 18:57:12 ok Oct 18 18:57:31 slyon_: I think i cannot attend or ill take my son and leave at 7.30pm Oct 18 18:59:30 nschle86, ok. no problem Oct 18 19:00:44 slyon_: you can send me a mail when you know the location and time so i can manage to come for 1-2 hours Oct 18 19:02:36 nschle86, alright. will do Oct 18 19:03:10 slyon_ Oct 18 19:03:27 slyon_: temporary bye, i need to reboot Oct 18 19:04:38 bye Oct 18 19:24:24 I've to leave. bye! **** ENDING LOGGING AT Fri Oct 19 03:00:01 2012