**** BEGIN LOGGING AT Mon Mar 14 02:59:58 2016 Mar 14 09:39:42 if i "addtask" a function A after another task B in my recipe, does it automatically imply that A should always run whenever B runs, or does it only setup the dependency? i'd like to only run A when specified through "bitbake -c A myrecipe", but for some reason it runs as part of "bitbake myrecipe" Mar 14 09:47:08 (i had a look at bitbake -g, and yes, it does seem like A is a dependency to do_build. anyway I can prevent that?) Mar 14 09:47:30 i basically want something like populate_sdk, a task that I can run manually but isn't part of the normal do_build-thing Mar 14 11:06:07 mago_: if you don't say "before do_build" (or indeed any other task) then it won't be part of the "normal" task flow Mar 14 11:36:03 Anyone familiar with having multiple partitions for the generated filesystem image (beyond just boot, rootfs, and swap)? Mar 14 12:02:30 I'm just looking for options. I've tried using wic by adding extra "part" directives to my wks file with different paths, but somehow the size always seems to go beyond the size of my SD card. Mar 14 16:03:27 eengie: two things come to mind, one, the rootfs size is controlled by separate variables, and there's one to add 'extra' space to the rootfs, which would bloat it up if it's more than you need/want, and two, partition alignment could be a factor Mar 14 16:06:06 kergoth: I ended up removing the --size option in my two rootfs parts only to realize that the rootfs plugin is going to grab everything at / and mount it at the path... not get everything at /path and mount it at /path which wasn't really intuitive to me. Mar 14 16:06:47 By removing the size option, the partitions shrank to a minimal size and things fit, but I found that /whatever partition would end up being the root file system, again. Mar 14 16:06:48 ah, yes, i can see how that would be confusing. perhaps we need a 'source path' vs 'target path' to distinguish Mar 14 16:06:56 or just change the behavior altogether Mar 14 16:07:11 we do need to make it easy to break up the rootfs into multiple partitions by path, if that doesn't work, that's a problem Mar 14 16:07:16 The format of the file made me immediately think of an fstab entry. Mar 14 16:07:27 So I had that experience working against me I think... Mar 14 16:07:45 i think that's fair, i expect that will be the case for a lot of folks Mar 14 16:08:05 Right, I don't think that works. From what I've read on forums, people want that feature and have proposed various patches, but I don't have access to any of that on daisy. Mar 14 16:08:12 i could see an argument for changing te default behavior, then adding a new argument to force the source path to be separate from the dest path Mar 14 16:08:15 i.e. --source-path / Mar 14 16:08:24 ah Mar 14 16:08:36 Right, that was one suggestion. Another was having it sourced from bitbake variables, which makes more sense. Mar 14 16:08:45 i'm not a wic expert by any means, just use it from time to time, haven't had time to try out the partitioning to break up the rootfs yet Mar 14 16:08:49 My route was to have a script that copied my WKS and used sed to inject path names for partitions. Mar 14 16:08:59 i want to play with it to experiment with read-only-rootfs handling without using volatile storage Mar 14 16:10:25 the wic guys tend not to hang out here :( Mar 14 16:10:36 that's unfortunate Mar 14 16:15:32 or we need to get more people using it :) Mar 14 16:17:14 at mentor we're using it (wic via the image type) for all our supported bsps now, with success, though we're hitting a hiccup now since live/bootdisk setups run after do_image in a separate task, they don't integrate well, have to run wic manually for that one for now, or add another task Mar 14 16:20:35 hi everyone, i installed mariadb but i cannot acces remotely, How can i fix this Mar 14 16:32:18 LinuxMice: I'm not all that familiar with mariadb, but I know that for MySQL I had to establish external users in the database and give them access to the table(s) of interest. Mar 14 16:56:11 kergoth: Is there a way to determine the family of processor being built by a given machine spec? (armv7l vs. x86, etc.)? Mar 14 16:57:52 look at DEFAULTTUNE, TUNE_FEATURES, and TARGET_ARCH in bitbake -e Mar 14 17:02:07 thanks Mar 14 18:02:30 Does anyone use FILESEXTRAPATHS_prepend to merge patch directories? Mar 14 18:03:27 "merge patch directories"? Mar 14 18:04:42 I have 5 recipes that all have the same patch, but then each has their own individual patches too Mar 14 18:05:12 filespath could certainly be used to deal with that, yes Mar 14 18:06:17 So that was a generic example; with another recipe it uses one of the common files but implements its own version of the other. Mar 14 18:06:45 When the prepend or append happens, it seems to ignore the one in the recipe's files folder. Mar 14 18:07:35 I would have thought that the combination of FILESEXETRAPATHS would have basically created a union of the two folders. Mar 14 18:08:11 with the last one defined taking precedence... Mar 14 18:09:28 prepend/append is cumulative, and doesnt'r emove anything Mar 14 18:10:16 I guess I was expecting a "cp -rf" kind of behavior where if the file exists, the next attempt to copy that file replaces the previously copied one. Mar 14 18:10:18 and so on Mar 14 18:15:44 it's like PATH. the first file found is used Mar 14 18:16:17 when a relative file:// file is specified in SRC_URI, FILESPATH is searched like PATH to find it Mar 14 18:16:19 I see... but probably the reason why append didn't work is because I was using :=. Mar 14 18:16:23 FILESEXTRAPATHS is just injected into FILESPATH Mar 14 18:17:35 := will force immediate expansion, which can be good or bad depending on what you're doing. if you're using FILE_DIRNAME or THISDIR in a bbappend, you have to use := to make sure it's relative to the append (file currently being parsed) rather than when it's expanded at the end, at which point it'd be the path to the recipe, not the append Mar 14 18:17:45 if you're using it in a recipe, there's no need for the immediate expansion Mar 14 18:24:49 Thanks for the explanation. My situation was from a recipe, but perhaps this has to do with where FILESEXTRAPATHS gets injected into FILESPATH, because if it's always before the local /files folder, then that explains why _append didn't affect the ordering of which file was ultimately applied. Mar 14 18:26:21 it sounds like you might want to append directly to filespath rather than filesextrapath Mar 14 18:26:30 but just use bitbake -e yourrecipe to examine the values of FILESPATH and FILESEXTRAPATHS Mar 14 18:26:40 bitbake -e is always the best way to make sure things are set the way you think they are Mar 14 18:26:54 Good point... Should have gone there first. Mar 14 18:27:12 Nice that it's granular enough to inspect a specific recipe (I've been using it for the whole image) Mar 14 18:27:36 no, you've been using it to examine configuration metadata, which is global Mar 14 18:27:40 the image is just another recipe :) Mar 14 23:08:04 http://pastebin.com/vBrXi92t did anyone see this already when dealing with extensible sdks? Mar 15 01:06:37 fischerm: I hadn't... I presume this is on reasonably recent master? Mar 15 01:23:07 bluelightning, yeah Mar 15 01:23:23 fischerm: did you need to do anything special to trigger that? Mar 15 01:23:32 bluelightning, just install it Mar 15 01:25:52 hmm, let me try that now Mar 15 01:26:29 I gotta run, I'll be back tomorrow am Mar 15 01:26:32 :-/ **** ENDING LOGGING AT Tue Mar 15 02:59:58 2016