**** BEGIN LOGGING AT Wed Dec 11 02:59:59 2013 Dec 11 05:04:06 hi can i know how to clean all my packages Dec 11 05:04:26 i used bitbake -c cleanall ? Is this correct ? Dec 11 06:22:09 santu: i only see clean and cleansstate in my image tasks Dec 11 07:34:41 santu: you can't clean all packages at once. bitbake -c cleanall will clean the recipe WORKDIR, remove the sstate content from that recipe and remove any downloaded file for that recipe. Dec 11 07:35:20 if you really want to clean all pacakges, deleting the TMPDIR folder is probably what you have to do. Dec 11 08:34:02 hi koen, so i finally got some time to look into u-boot (from meta-beagle) not working well for deploy/sstate. i realized that we are directly hacking in DEPLOY_DIR_IMAGE in uboot do_deploy, which we should work in DEPLOYDIR. Dec 11 08:34:16 as such the sstate is empty for deploy . Dec 11 08:34:26 the following patch seems to be doing the right thing: http://pastebin.com/ap2kzE7X Dec 11 08:35:15 when i look at ./48/sstate-u-boot-denx-beaglebone-oe-linux-gnueabi-2013.10-r0-beaglebone-3-48876e024325fc3a4b1b89f36918261f_deploy.tgz in my sstate after and before, the file is no longer empty Dec 11 08:35:56 i checked a couple of other recipes (in oe-core and in meta-ti) and they all work out of DEPLOYDIR. Dec 11 08:36:19 i can prepare proper patches for that, but wanted to run that by you quickly. Dec 11 08:37:19 that's weird Dec 11 08:37:27 since deploydir used to be foo/deploy Dec 11 08:37:35 and deploy_dir_image foo/deploy/images Dec 11 08:37:43 (and now foo/deploy/images/$MACHINE) Dec 11 08:39:37 koen: hmm. no. DEPLOY_DIR_IMAGE is the final deploy dir, while DEPLOYDIR is local, e.g. in WORKDIR. Dec 11 08:39:44 ah Dec 11 08:39:51 that makes sense Dec 11 08:40:05 which list should i send the patches to? Dec 11 08:41:03 dunno, for meta-beagle you can email me directly Dec 11 08:41:15 koen: ok... ;-) i will send 2 series for dylan and master Dec 11 08:41:19 thanks! Dec 11 08:41:24 since the filename has changed in between ;-) Dec 11 08:41:31 yes Dec 11 08:41:40 I've screwed up twice with cherry-pick already Dec 11 08:41:53 that's more or less expected... Dec 11 08:42:21 these grub recipes make my eyes bleed Dec 11 08:42:37 I'm tempted to just write a grub_git.bb from scratch Dec 11 09:20:58 koen: have you already klibc_2.0.3? Otherwise I'll update in a couple of days. Dec 11 09:32:53 no, I only did 2.0.2 Dec 11 09:34:05 [koen@rrmbp grub]$ git log --oneline 2.00 | wc -l Dec 11 09:34:05 7027 Dec 11 09:34:11 *sigh* Dec 11 09:34:41 morning all Dec 11 09:34:45 hey bluelightning Dec 11 09:35:07 hi koen Dec 11 09:35:24 koen: did you know you can kexec grub? Dec 11 09:35:40 I didn't Dec 11 09:36:55 personally, I don't use grub on EFI machines Dec 11 09:37:25 but working on a new grub recipe will pay the rent, so there I go Dec 11 09:37:39 Anyone running webstuff in node.js? it uses a lot of cpu for simple tasks on my system. Wonder if I should go with lighttpd or if there are other tweaks to be done. Dec 11 09:38:01 define 'simple tasks' Dec 11 09:38:21 lighttpds won't run code, so I guess you mean lighttpd+fastcgi ? Dec 11 09:39:04 koen: I guess that too. know little and haven't researched. Dec 11 09:40:03 koen: it uses 60-90% cpu load for 3-5 seconds when I log in. Dec 11 09:40:43 it's javascript, you know... Dec 11 09:42:07 I even don't know how your js code would work with fcgi :) Dec 11 09:42:56 ynezz: I know next to nothing about webdevelopment, which is why I started out with node.js in the first place :) Dec 11 09:43:54 and you've seen any serious stuff done with it? Dec 11 09:49:13 ynezz: no. I see they have a braglist on their site about who uses it, but I don't know *how* they use it. Dec 11 09:49:54 chose it because I didn't know web and had to make something working fast. it has served me well, but now I need to speed up things. Dec 11 09:50:03 tasslehoff: is that a one-time startup hit, or every request? Dec 11 09:50:15 tasslehoff: tbh, you're better off in a nodejs channel Dec 11 10:01:25 rburton: every request. I'm in nodejs as well. this was more to get an embedded view of things :) Dec 11 10:37:09 good morning Dec 11 10:37:52 how can I monitor the do_fetch phase progress during a bitbake? Dec 11 10:59:59 mckoan: tail -f on log.do_fetch? Dec 11 11:00:05 I don't think there's another way Dec 11 11:00:56 there is a bug open for more useful progress indication: https://bugzilla.yoctoproject.org/show_bug.cgi?id=5383 Dec 11 11:03:25 bluelightning: very nice proposal Dec 11 11:03:54 actually I have to wait a loooong timeout before understand it failed Dec 11 11:04:54 it probably wouldn't be too hard to implement; the key issues are (a) getting the progress info out of whatever we're using to do the fetch (git, wget, etc.) and (b) forwarding that to the bitbake UI in the form of events Dec 11 11:38:24 is it possible to skip a do_fetch? Dec 11 12:05:45 mckoan: well, if you must do it, if you have the file in DL_DIR just touch .done next to it and it won't actually do anything during do_fetch Dec 11 12:45:29 hmmm Dec 11 12:45:30 /build/linaro/build/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib/rpm/bin/debugedit: /build/linaro/build/build/tmp-eglibc/work/aarch64-oe-linux/grub/2.00+AUTOINC+16c8f78513-r0/package/usr/lib/grub/arm64-efi/gcry_dsa.module: Unhandled relocation 258 in .debug_info section Dec 11 12:45:34 ah well, it builds Dec 11 12:46:44 Hi, can someone tell me how to add existing user to a group? Dec 11 12:49:14 koen: is this with ld-is-gold? Dec 11 12:49:34 ensc|w_: not sure Dec 11 12:49:52 I did not had success to get a ld-is-gold built grub2 to run and had to apply dirty hacks to build it with the bfd linker Dec 11 12:50:08 I tried GROUPMEMS_PARAM, it is not working Dec 11 12:50:19 but that was 9 months or so ago and things might hhave changed in the meantime Dec 11 13:09:25 in which meta-layer can I find plymouth? layers.openembedded.org gives me no matches Dec 11 13:13:29 maybe the it's not packaged anywhere Dec 11 13:14:32 rburton: yes thanks, i use psplash - comes from the yocto-peoples Dec 11 13:33:04 hey, are .scc (and others kernel meta) yocto-specific (or supposed to be) ? Dec 11 13:49:20 bencoh: I think scc files may be specific to the linux-yocto style of kernel configuration yes Dec 11 14:12:08 Is there any way to use two git repositories in single SRC_URI in a recipe Dec 11 14:12:10 ? Dec 11 14:28:46 Hello everybody. I would like to ask a little off-topic question. I want that the next 'opkg upgrade' command will install a new package that is not really dependant to other packages. Is that possible ? Dec 11 14:49:42 otavio, thanks for taking my patches Dec 11 15:12:51 bluelightning: looks like they get included/generated (?!) when building a linux-yocto kernel Dec 11 15:13:59 vadmeste: don't think you can, unless you introduce a dependency, or use RREPLACES+RPROVIDES+RCONFLICTS to replace an existing package Dec 11 16:26:36 Crofton|work: you're welcome Dec 11 16:26:49 dv_: please send the fix for the mailing list Dec 11 17:58:30 what's the proper way to punt dropbear in favor of openssh? but in oe-classic.. Dec 11 18:25:35 As we continue our product level testing of our usages of OE.. we're running into problems where scripts are using specific "standard" system tools, and they're not listed in the RDEPENDS... Dec 11 18:25:52 i.e. /usr/bin/env.. any python script pretty much uses it.. and either coreutils or busybox can provide it.. really doesn't matter which.. Dec 11 18:26:15 RPM is smart enough to automatically inser a dependency to /usr/bin/env -- but bitbake has no idea that dependency exists.. Dec 11 18:26:29 so you can end up with a system install w/o /usr/bin/env, causing failure(s).. Dec 11 18:26:39 similarly start-stop-daemon with the initscript scripts has hit us.. Dec 11 18:27:20 One possible answer is add RPROVIDES/RDEPENDS of say "virtual/usrbinenv" or "virtual/start-stop-daemon".. etc.. Dec 11 18:29:57 is there any suggestion, or do I need to just move this to the mailing list/bugzilla? Dec 11 18:30:32 I don't want to 'invent' something locally and then find out it's the wrong way to do it, etc.. whatever we end up using for our system/customers.. I'd like to make sure the community as a whole is comfortable with.. Dec 11 18:32:36 fray: ultimately some form of dependency has to be stated up front Dec 11 18:33:24 otherwise there's no way for bitbake to factor that dependency into which things need to be built Dec 11 18:33:49 (by up front, I mean in the recipe) Dec 11 18:35:15 I suspect this gets even more tricky when you start removing things from busybox's configuration; we don't have any way of handling when something "standard" from there is implicitly expected by some recipe and ends up not being provided Dec 11 18:51:31 bluelightning ya, I know.. in many cases things like the usrbinenv and start-stop-daemon (in busybox) can be determined by the configuration early.... Dec 11 18:59:19 anybody else even using oe-classic? Dec 11 19:00:25 there are people Dec 11 19:00:32 but they aer slowly switching over Dec 11 19:00:43 mostly legacy products Dec 11 19:01:38 any advice on the openssh-server replacing dropbear thing? Dec 11 19:20:30 nm, found it Dec 11 19:20:46 needs DISTRO_SSH_DAEMON set instead Dec 11 21:52:25 hmm totem isn't available anywhere in OE, right? Dec 11 22:09:47 dv_: there's an old recipe in OE-Classic that could presumably be brought up-to-date Dec 12 01:20:25 JaMa, awesome job finding interesting corner cases Dec 12 01:22:19 hehe doing a lot of builds with weird metadata does that for me :) Dec 12 01:23:07 found 5 more similar cases in private layers **** ENDING LOGGING AT Thu Dec 12 02:59:59 2013