**** BEGIN LOGGING AT Thu Jun 16 02:59:58 2016 Jun 16 07:04:05 morning Jun 16 09:07:59 lo! is there an update of glibmm planned? as the latest version in meta-oe is full of ugly warnings when it's used with the regular g++ version that comes with yocto... Jun 16 15:22:28 what is the best way to prevent /var/log from being symlinked to /var/volatile/log ? Jun 16 15:44:20 So, I always forget about bitbake's options to load other config files, but it was quite handy for this situation where i had to temporarily scale down the threads/jobs for certain tasks to stop OOM killed tasks Jun 16 15:44:21 bitbake -R conf/slow.conf qtwebkit webkitgtk && bitbake -k world Jun 16 15:44:22 heh Jun 16 15:45:00 cbrake: hmm, there's the base-files link, possibly volatiles, possibly systemd tmpfiles config, and there's the fs-perms.txt that redirects packages including files in the volatile dest Jun 16 15:45:11 not sure what all would be involved :\ Jun 16 15:45:16 maybe someone else knows better Jun 16 15:45:27 kergoth: yeah, found that one. Now, in the nginx package, its also creating the link Jun 16 15:46:22 when working on r/o rootfs i kept running into recipes packaging files inside of volatile paths, quite a pain. yes, fs-perms redirects them to the dest of the link, but then they just end up in the path where the tmpfs mounts over, so all those needed fixing to use volatiles/tmpfiles :\ Jun 16 15:46:30 cbrake: heh, that sounds like a bug in the recipe Jun 16 15:46:35 unless it's fs-perms doing it Jun 16 15:47:21 [cbrake@mars 1.4.4-r2]$ find -name log Jun 16 15:47:21 ./packages-split/nginx/var/log Jun 16 15:47:21 ./packages-split/nginx/var/volatile/log Jun 16 15:47:21 ./package/var/log Jun 16 15:47:21 ./package/var/volatile/log Jun 16 15:47:22 ./image/var/log Jun 16 15:47:28 odd Jun 16 15:47:31 wonder if there is something in the packaging code Jun 16 15:47:37 that automatically does it Jun 16 15:47:49 see fs-perms.txt :) Jun 16 15:47:57 * cbrake looks Jun 16 15:48:06 specifically meta/files/fs-perms.txt Jun 16 15:48:20 it corrects permissions/etc on packaged files, but does more than that when it comes to those links Jun 16 15:48:26 affects the packaging of everything, afaik Jun 16 15:48:44 afaik, anyway, could be remembering wrong Jun 16 15:48:59 there it is, ${localstatedir}/log link volatile/log Jun 16 15:49:17 i'd try removing that nad see if it change sthe behavior Jun 16 15:49:28 trying ... Jun 16 15:50:59 that is better: Jun 16 15:51:02 [cbrake@mars 1.4.4-r2]$ find -name log Jun 16 15:51:03 ./packages-split/nginx/var/log Jun 16 15:51:03 ./package/var/log Jun 16 15:51:03 ./image/var/log Jun 16 15:51:06 might have to force a re-run of do_package, unless file-checksums is set for it Jun 16 15:51:09 nice Jun 16 15:51:24 now, how to use a custom fs-perms.txt ... Jun 16 15:51:48 i'm guessing it looks for 'files/fs-perms.txt' in BBPATH, so another layer could override it if it's earlier in bblayers, but i've never tried Jun 16 15:51:59 would likely be nice if it wasn't one monolithic thing.. Jun 16 16:04:32 yes, must moved it to files in layer and that work -- thanks for the help Jun 16 16:08:04 np Jun 16 16:23:06 cbrake: you can set custom fs-perms via FILESYSTEM_PERMS_TABLES Jun 16 16:23:34 khem: nice Jun 16 16:24:20 ah, sweet, thanks khem Jun 16 16:24:37 the question is, does that let you *remove* something done by fs-perms.txt, or only add? Jun 16 16:38:58 if you set it then default fs-perms.txt is not used at all Jun 16 16:39:07 ah, right Jun 16 16:53:15 I just pulled oe master and I'm getting a bb.BBHandledException: invalid token in terminal.bbclass, line 32... Jun 16 16:53:18 is this a known issue? Jun 16 16:53:41 wait.. I'm not on master.. Jun 16 16:53:43 never mind.. sorry Jun 16 17:34:51 <3 wget-finder Jun 16 19:55:26 hmm, bitabke keeps saying: NOTE: consider defining a PREFERRED_RPROVIDER entry to match java2-runtime Jun 16 19:55:30 even though I have that defined Jun 16 20:04:45 bitbake sure loves re-running setscenes that it already ran a build or two ago, for no apparent reason Jun 16 20:04:54 typo? /me does that often Jun 16 20:08:09 on an unrelated note, i wish more layers that provide multiple providers of something would arrange for a default, either through DEFAULT_PREFERENCE or ??= preferences via layer.conf Jun 16 20:08:24 irritating tha tincluding a given layer means you get warnings unless you set the prefs yourself Jun 16 20:13:17 paulg_: I will follow up the vte email with a potential fix, having the old vte version live side-by with the new Jun 16 20:13:52 that should allow xfce4-terminal to build, will see if others will need to point to the old lib Jun 16 20:14:57 sounds reasonable. Jun 16 20:15:19 kergoth: unfortunately DEFAULT_PREFERENCE doesn't work across layers, so you're right if there are multiple providers in the same layer it makes sense to set DEFAULT_PREFERENCE but to provide alternative version to something which lives in some other layer, DEFAULT_PREFERENCE doesn't work Jun 16 20:15:32 ah, that's unfortunate :\ Jun 16 20:15:51 would be nice if there was a good solution for that Jun 16 20:15:54 hmm Jun 16 20:16:12 https://bugzilla.yoctoproject.org/show_bug.cgi?id=2964 <- not a bug.. :/ Jun 16 20:19:26 if DEFAULT_PREFERENCE is working as designed, then we need a different solution. there has to be a way for a layer to add a weak provider for something which isn't selected by default and doesn't warn the user if they lack a preference Jun 16 20:28:50 DEFAULT_PREFERENCE I thought should act globally Jun 16 20:37:57 yeah, i'd expect it to exert more control than layer priority does, but less than preferred_provider, personally... Jun 16 20:37:58 * kergoth shrugs Jun 16 20:44:48 sometimes an older recipe from higher priority layer gets chosen Jun 16 20:44:54 thats a problem too Jun 16 20:52:48 agreed, folks often find that confusing Jun 16 20:53:26 agreed with both Jun 16 20:59:17 perhaps time we discussed this on the oe-architecture list **** ENDING LOGGING AT Fri Jun 17 02:59:58 2016