**** BEGIN LOGGING AT Fri Jan 22 02:59:58 2016 Jan 22 08:21:31 good morning Jan 22 12:28:20 Hi guys, I've just found this spdx.bbclass, and I'd like to know how can I use it: http://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/meta/classes/spdx.bbclass Jan 22 12:48:56 Hello, I'm trying to follow http://www.jumpnowtek.com/yocto/Add-Fortran-support-to-a-Yocto-build.html to add fortran on fido, but I got an issue with gtrh-default.h which is missing compiling libgfotran. Is there a workaround ? Jan 22 16:31:52 Hello, I managed to compile libgfortran on jethro, but I have this error at the very end "QA Issue: gcc-runtime: Files/directories were installed but not shipped in any package". Any way to solve it ? Jan 22 16:32:23 minipada: check ${WORKDIR}/image and see which files are not in ${WORKDIR}/packages-split. Add them to a FILES_* variable. Jan 22 16:33:01 Or was that ${WORKDIR}/package for the first one Jan 22 16:33:56 it seems it is package for the second Jan 22 16:34:31 so in my case, it would be FILES_gcc-runtime+= ... Jan 22 16:34:31 ? Jan 22 16:39:09 more likely FILES_${PN}, otherwise the existing FILES_${PN} will overwrite your FILES_gcc-runtime, since PN is gcc-runtime. Jan 22 16:39:25 also the latter wouldnt' be used at all for e.g. nativesdk and other variants Jan 22 16:39:33 would also break with multilib Jan 22 16:42:04 mmh ok, it makes sense, I'm trying Jan 22 16:42:18 Thanks for your fast answers Jan 22 16:49:20 OK, still not good. I tried in the .inc and .bb Jan 22 16:57:31 'not good' isn't of much use. did you use bitbake -e to make sure the variable is set to what you think it is? Jan 22 16:57:38 that'd be a good first step, check our assumptions Jan 22 16:58:03 ok, didn't know about that Jan 22 17:08:46 I checked, with bitbake -e, not sure how to debug, I can read "FILES_gcc-runtime=" (and the files I needed), still the same error Jan 22 18:21:17 gm all Jan 22 18:23:58 good morning :) Jan 22 18:24:23 Just realized there's a _remove today while working on a bbappend... huzzah! Jan 22 18:29:14 jynik: yes its handu Jan 22 18:29:16 handy Jan 22 18:29:45 its always good to look through documentation. I always find a lot of useful things everytime I look into it Jan 22 18:30:25 fair warning, _remove cannot be undone. a later _append will not counteract an earlier _remove Jan 22 18:30:35 meaning if you use _remove in a bbappend, a higher priority layer can't override what you've done Jan 22 18:30:38 so it doesnt' play well with others Jan 22 18:30:48 i don't recommend its use in bbappends for that reason Jan 22 18:35:10 kergoth: Thanks for the warning. Jan 22 18:36:03 Usecase here was that I added a bbappend to bring linux-firmware up to master for some newer ucode Jan 22 18:36:35 but one of the licenses in the LIC_FILE_CHKSUM no longer exists... so I /think/ I won't be shooting myself in the foot? Jan 22 18:36:57 I guess we'll see how long until I put said foot in my mouth... Jan 22 18:37:17 Definitely open to better solutions to this. Jan 22 18:39:05 khem: Yeah, the Yocto project team does a fantastic documentation job imho Jan 22 18:39:27 I always keep the "megamanual" a click away :) Jan 22 18:41:30 jynik: you could just write a new recipe in your own layer Jan 22 18:41:37 and if relevant upstream it as well Jan 22 18:49:56 khem: The "correct" solution would be for me to submit a patch for the linux-firmware_git I believe. Jan 22 18:50:05 yes Jan 22 18:50:20 However, I'd like to stick on HEAD of the latest release in my org Jan 22 18:50:35 So would something like that get backported to jethro? I imagine not, right? Jan 22 18:59:04 jynik: it can if you justify it and provide reasons Jan 22 19:05:39 khem: Alright. In this case, I'd so there's absolutely no justification. I just happen to be on Jethro and using an newer kernel (via my own recipe) to bring up some HW... Jan 22 19:05:55 s/I'd so/I'd say/ Jan 22 19:09:35 So... I guess I'll be a good neighbor and see if it's worth tossing upstream Jan 22 19:09:52 Nice to have things around after you update ;) Jan 22 19:57:41 Is it possible to add just a single entry from a recipes "PACKAGES" to INSTALL_append? Jan 22 19:58:26 Along the lines of the linux-firmware recipe, it describes packages for specific devices; only one of which I'm concerned with. Jan 22 19:59:26 *IMAGE_INSTALL_append, rather Jan 22 20:00:00 rp, thanks for going on stackoverflow Jan 22 20:00:11 I just upvoted one of your answers Jan 22 20:00:55 upvoted rburton Jan 22 20:12:18 Crofton|work: stackoverflow do we have yocto or oe label Jan 22 20:12:51 khem, yes Jan 22 20:13:02 need to search on OpenEmbedded, yocto and poky Jan 22 20:13:21 oh cool. I was suggesting it in 2015 ELC AB meeting Jan 22 20:13:52 manpower is issue Jan 22 20:14:00 look like rburton has been doig some Jan 22 20:14:05 Anders is also good about it Jan 22 20:22:39 Crofton|work: I added bitbake yocto and openembedded tags to my profile Jan 22 20:22:48 Crofton|work: I wil answer in future Jan 22 20:22:51 thanks Jan 22 20:23:01 I think I screwed up and two account Jan 22 20:23:17 may have answered one with link to getting started and some guy bitched about links since they die Jan 22 20:23:25 can't find it now :( Jan 22 20:23:29 heh Jan 22 20:23:40 I get angry at so answers that bitrot :) Jan 22 20:24:16 I need to talk to Jefro to see if there is anyway we can get stats from SO Jan 22 20:54:57 anyone know the uclibc in here? I get an error configuring gcc-runtime: checking dynamic linker characteristics... configure: error: Link tests are not allowed after GCC_NO_EXECUTABLES. Jan 22 21:09:56 tripzero2: are you using master Jan 22 21:12:05 khem: possibly not. does it work on master? Jan 22 21:12:20 master poky? master uclibc? Jan 22 21:12:41 master of poky Jan 22 21:12:46 it should work Jan 22 21:12:59 i'll try Jan 22 21:14:43 hmm there is a big pull request due from me which has some uclibc fixes too Jan 22 21:15:01 ooo, that's exciting Jan 22 21:15:06 my tree is here Jan 22 21:17:28 https://github.com/kraj/openembedded-core/tree/kraj/master Jan 22 21:19:18 looks like pulling a more recent poky fixes that error Jan 22 21:19:37 khem, if I run into more, i'll try your tree Jan 22 21:21:17 tripzero2: ok. Jan 22 21:21:51 actually, After the pull, enough needs to be rebuilt that I can't say for sure anything was fixed... :P Jan 22 21:26:34 thats what I thought Jan 22 21:26:42 but machines could be fast these days Jan 22 21:55:31 khem: no dice. I'll try master on your tree. Jan 22 22:52:48 Crofton|work: i've a search in my rss feed, so pop along and answer/correct where i can Jan 22 23:12:01 Crofton|work: meant to look at stackoverflow for a while. It is a good way to find out the questions users have, and try and get good quality answers to them Jan 22 23:12:34 the headache is building reputation Jan 22 23:20:04 Crofton|work: I seem to have the basic rights in a couple of days... Jan 22 23:20:49 It helps if you do some of their exercises. I can finally vote on answers. Jan 22 23:21:49 we should have a mutual appreciation society, someone ask a question, everyone else answer, everyone upvote everyone else Jan 22 23:21:51 and rotate Jan 22 23:21:56 +1 Jan 22 23:22:32 maybe we should have a stack overflow sprint one "day" Jan 22 23:22:40 as in a 1-2 hour block of time Jan 22 23:56:56 Crofton|work: I did have a look through the unanswered questions and put in a few Jan 22 23:57:08 rburton: +1 on the rotation :) Jan 23 00:03:57 kergoth: I tried your suggestions for wic Jan 23 00:04:48 it errors out like https://gist.github.com/ab6545619496f8d02e07 Jan 23 00:06:16 right outta here Jan 23 00:06:20 have a good weekend all Jan 23 00:09:12 RP: whats your take on musl for poky-tiny ? Jan 23 00:12:53 khem: probably in favour. I see there is a patch and it seems to have worked well Jan 23 00:14:49 RP: yes. I have plans to promote poky tiny in broadband community and they only know openwrt Jan 23 00:15:11 khem: would likely do it good to get some attention :) Jan 23 00:15:24 khem: pretty sure there is more that could be some space wise there too Jan 23 00:16:14 I think we should tryin some low hanging fruits like using -Os for global opts Jan 23 00:16:34 khem: just wish I had the time :) Jan 23 00:16:44 I am going to try that Jan 23 00:16:52 khem: sounds good to me Jan 23 00:17:22 I sent a patch earlier today which lets us use busybox all the way to login Jan 23 00:18:16 650k for rootfs is quite a good deal I think Jan 23 00:18:29 khem: probably needs to go into local.conf.sample.extended btw Jan 23 00:18:38 khem: we try and keep the base sample lean Jan 23 00:19:06 I thought so but then resorted to local.conf Jan 23 00:19:14 but I think its extended option Jan 23 00:19:41 I think we can trim down busybox Jan 23 00:19:52 khem: for tiny, that would make sense Jan 23 00:20:02 yes for tiny Jan 23 00:21:45 RP: is poky tree 1 to 1 with oe-core now a days ? Jan 23 00:22:09 RP: thinking a case where a patch is done on poky will always apply directly on oe-core repo Jan 23 00:23:13 khem: there is some file masking but it is a straight merge of bitbake + oe-core + meta-yocto + docs Jan 23 00:23:34 khem: local.conf.sample is in a different location in both but I keep them pretty much in sync Jan 23 00:24:19 RP: I see Jan 23 00:24:35 I created a github mirror of poky btw Jan 23 00:24:50 https://github.com/kraj/poky Jan 23 00:25:17 khem: ok. I really struggle with the github ui tbh, just doesn't work for me... Jan 23 00:25:26 I wanted to move my day to day dev on this but with some awareness of differences Jan 23 00:25:40 RP: I got used to it Jan 23 00:25:49 but its different Jan 23 00:26:05 khem: there are no actual code differences, just some files masked out (e.g. bitbake's bitbake.conf) Jan 23 00:26:16 and there aren't many of those Jan 23 00:27:04 so poky(poky-tiny)+meta-openwrt will do it for these folks next week Jan 23 00:27:19 khem: sounds good :) Jan 23 00:27:30 openwrt has its own init system called procd Jan 23 00:27:41 its quite interesting Jan 23 00:27:50 very small can boot in parallel Jan 23 00:28:05 sounds interesting Jan 23 00:28:14 and a IPC bus implementation which is not dbus Jan 23 00:28:21 its called ubus Jan 23 00:28:43 I do have to wonder what was wrong with dbus for that :) Jan 23 00:28:52 size Jan 23 00:29:03 and performance Jan 23 00:29:23 we tried like 2000 simultaneous voice calls Jan 23 00:29:27 and dbus sucks Jan 23 00:29:48 ubus works for that? Jan 23 00:30:18 it shows like 1000% improvement Jan 23 00:30:24 in latency Jan 23 00:31:00 static linking libdbus helps a bit Jan 23 00:31:13 I guess that wasn't really what dbus was designed for :) Jan 23 00:31:23 I guess yes Jan 23 00:31:26 heh Jan 23 02:13:36 kergoth: if I add a new line starting bootloader in .wks file then it works Jan 23 02:13:50 kergoth: it seems bootloader cmd is required too Jan 23 02:15:13 ah, makes sense Jan 23 02:15:33 but its more grub or syslinux oriented Jan 23 02:15:40 uboot, I am not sure Jan 23 02:16:29 so I took one of 8G configs and added bootloader to it Jan 23 02:16:46 put it in scripts/lib/wic/canned-wks/sdimage-bootpart-8g.wks Jan 23 02:16:56 then in local.conf Jan 23 02:16:59 added Jan 23 02:17:30 IMAGE_FSTYPES_append = " wic.bz2" Jan 23 02:17:36 IMAGE_DEPENDS_wic_append = " dosfstools-native mtools-native virtual/bootloader" Jan 23 02:17:42 WKS_FILE = "sdimage-bootpart-8g.wks" Jan 23 02:17:47 IMAGE_BOOT_FILES_beaglebone ?= "u-boot.img MLO" Jan 23 02:18:11 and now I get sdcard image Jan 23 02:18:16 how good it is dont know Jan 23 02:19:10 kergoth: I think sparse file which can be installed into arbitrary sized sdcards would be desired Jan 23 02:35:13 khem: I don't really see how that'd be possible given the fs inside the image has to have a fixed size. Jan 23 02:35:49 well, if you just use sdcard-bootpart, it'll be just teh size of the rootfs, tiny, and then you could grow the partition and the fs both at runtime Jan 23 02:36:00 i don't think sparse would help that much, since the fs and partition still need to be adjusted Jan 23 02:36:30 you can do a conv=sparse dd onto the disk, and avoid writing the whole thing, but then there's the risk remnants from previous flashes are still around **** ENDING LOGGING AT Sat Jan 23 02:59:58 2016