**** BEGIN LOGGING AT Sat Mar 21 02:59:57 2020 Mar 21 03:36:00 New news from stackoverflow: meta-oe of meta-openembedded vs meta of Poky Mar 21 08:06:46 New news from stackoverflow: What is the use of distro less yocto image and how to set it in local.conf file Mar 21 11:33:09 Are links created by EXTERNALSRC_SYMLINKS actually needed for external source builds or can I set it to nothing? Mar 21 16:47:12 RP: any hints on avoiding recheck ( git ls-remote ) when using AUTOREVs ? I would like to checkout once and then not check again the first checkout is good Mar 21 16:49:01 I thought BB_SRCREV_POLICY = "cache" should do it but seems its not doing it Mar 21 17:23:38 khem: that would have been my guess as well Mar 21 17:24:19 RP: I think I should work towards a workflow that does not involve AUTOREV Mar 21 17:25:09 khem: AUTOREV is hard on the system. Its probably been broken by some change as its not likely well tested in automated tests Mar 21 17:30:57 RP: if there was a policy that checks it out and leaves it at that would fullfil my current need Mar 21 17:34:43 khem: that is what cache is meant to do Mar 21 17:34:50 khem: so really we should fix it Mar 21 17:40:29 BB_SRCREV_POLICY="cache" has worked in the past, i've used it. been a while since i've tested it, though Mar 21 17:42:13 kergoth: I am testing master and seems git ls-remote threads are running during parse Mar 21 17:42:29 huh, sounds like we need extra unit tests for autorev Mar 21 17:42:53 kergoth: I'm not sure we have any :( Mar 21 17:43:15 * RP has spent nearly two days chasing a devtool bug which turns out to be in the remote datastore code :( Mar 21 17:43:30 even seeing the issue I don't quite know what to do about it Mar 21 17:43:41 i hate autorev when used with remote repos, but i could see using it with local repos as an alternative to externalsrc, since externalsrc still uses nostamp iirc (we should really fix that) Mar 21 17:43:46 ugh Mar 21 17:49:27 kergoth: I think I agree with you here Mar 21 17:49:55 kergoth: it means a two step process, which means every component first commits the code change to component Mar 21 17:50:11 then commits the metadata/recipe srcrev bump Mar 21 17:51:09 hmm, i rather like the idea of using repo/submodules/whatever for actual source repos for in-development components alongside the layers, point the recipes to that, and arrange to pick up the local changes possibly with anonymous python Mar 21 17:51:12 * kergoth ponders Mar 21 17:52:42 kergoth: that does not scale for large projecrs with 100+ repos Mar 21 17:54:27 with devtool do we have a way to checkout tip of tree with say devtool modify ... Mar 21 17:58:17 true Mar 21 17:58:28 devtool modify runs into the everything-is-nostamp issue though Mar 21 18:01:34 kergoth: thats ok for devs though Mar 21 18:01:51 they usually work in one or two components/recipes Mar 21 18:01:54 at a time Mar 21 18:02:48 true Mar 21 18:04:10 so thinking about what you suggested with git submod design, will it track changes to sourcecode of those components Mar 21 18:04:45 I want to get rid of AUTOREVs Mar 21 18:08:28 * RP thinks the remotedata code is totally flawed :( Mar 21 18:08:29 New news from stackoverflow: How to build a linux for Intel galileo gen 2? Mar 21 18:08:42 and should it use file:// fetcher ? if so then will it copy sourcecode again into tmp/ Mar 21 18:08:47 during build Mar 21 18:21:03 hi guys, anyone has examples for a gitlab-ci based pipeline? I am currently using this one: https://gist.github.com/btgoodwin/028c3a63315a27cae3c5989455058eef Mar 21 18:29:09 kriive:that seems to be good start Mar 21 18:29:38 RP: with BB_SRCREV_POLICY = "cache" should I expect it to not reach out to check for revisions on second run ? Mar 21 18:30:11 or will it just mean it wont bump the HEAD but still do all git network operations like ls-remote etc. Mar 21 18:46:18 I add DEPENDS += "rsync-native" to build-applience recipe, but when I run my rsync action after do_fetch it says "command not found". Mar 21 20:06:10 hello I installed 'mesa' but a specific header is missing when Im compiling my app recipe , any idea how to debug ? Mar 21 20:31:59 dv: it has to be in HOSTTOOLS as well Mar 21 20:32:22 mauz555: are you using some SOC specific OpenGL implementation ? Mar 21 20:32:32 if so you might want to check it there Mar 21 20:38:53 khem: yes there is 'mesa' files from poky:zeus and small bbappend from meta-freescale Mar 21 20:41:58 any idea how and where to set a PACKAGECONFIG flag for a specific recipe? Mar 21 20:47:12 khem: ty I found the fix Mar 21 21:07:00 RP: I think this commit changed the behaviour https://git.openembedded.org/bitbake/commit/?id=ba093a38539960e645e994a66ed7872a604c00a9 Mar 21 22:23:17 khem: hmm, that does sound like its broken the correct behaviour :( Mar 21 22:24:17 RP: http://sprunge.us/64lmJh this fixes it Mar 21 22:24:21 for me Mar 21 22:24:42 khem: We should probably send email and ask Saur about it Mar 21 22:24:56 This is why we need more tests Mar 21 22:25:24 I have replied to original email but it does sound to me that its breaking existing behaviour Mar 21 22:26:33 khem: yes, its incorrect and I shouldn't have merged it Mar 21 22:27:10 should I send this patch on top to fix it Mar 21 22:52:13 khem: I don't understand that original commit, should it be reverted entirely? Mar 21 22:57:40 RP: yeah perhaps it should be reverted completely **** ENDING LOGGING AT Sun Mar 22 02:59:58 2020