**** BEGIN LOGGING AT Wed Oct 22 03:00:01 2014 Oct 22 06:56:17 hi, how can i tell oe where to get the makefile of a progect : sources are in src/ dir and makefile in parent dir Oct 22 07:04:16 ccd Oct 22 07:56:54 good morning Oct 22 09:33:52 bluelightning, how do I remove a layer fro mthe index? Oct 22 09:34:07 and I need to check on FOSDEM stand application Oct 22 09:52:30 Crofton|work: still haven't submitted Oct 22 09:52:43 Crofton|work: should be able to do it by end of week Oct 22 09:52:52 I fixed my layer depends for belen you can submit now :) Oct 22 09:53:09 Crofton|work: you can't, but I can if you let me know what you want removed Oct 22 09:53:18 I need to remove my old meta-zynq layer from the layer index Oct 22 09:54:32 ok, I'll delete it now Oct 22 09:54:38 thanks Oct 22 09:55:01 ok, it's gone Oct 22 09:56:26 Is there any way I can build OE on an ecryptfs filesystem? Oct 22 09:56:37 lag: no Oct 22 09:56:49 ecyptfs has trouble with long/deep paths Oct 22 09:57:05 well Oct 22 09:57:05 koen: That's pants! Oct 22 09:57:23 you could create a loopfile system, format that as ext4, etc Oct 22 09:57:33 koen: So anyone that builds OE has unencrypted /home directories? Oct 22 09:57:45 I don't build in /home Oct 22 09:57:55 I have /build Oct 22 09:58:17 but e.g. dm-crypt works Oct 22 09:58:30 * lag doesn't have enough disk space to create a new partition Oct 22 09:59:54 koen: How much disk space does a typical build take? Oct 22 10:00:32 the builds I do take >100GB Oct 22 10:00:41 but small builds fit in a 16GB tmpfs Oct 22 10:00:48 so something between that Oct 22 10:03:53 koen: Thanks Oct 22 10:06:11 koen: How long is this going to take if I build on an external USB 2.0 HDD? Oct 22 10:06:31 koen: Will I be able to have a USB 3.0 HDD delivered in the same time-frame? ;) Oct 22 10:15:42 small build is ~3 hours on a laptop (including fetching everything) Oct 22 10:16:03 I'd guess the usb2 hdd would add a few hours Oct 22 10:16:11 depending on your internet connection Oct 22 10:16:48 I've seen git take 4 hours to download linux-yocto due to sucky internet Oct 22 10:22:47 koen: I'll give it a go - ta Oct 22 10:24:50 Worst case you can always stop the build when the new disk turns up :) Oct 22 10:27:42 lag: FWIW our firm like many requires encryption, but we use dm-crypt Oct 22 10:27:46 much less evil... Oct 22 10:28:27 honestly the world would be a much better place had distro installers supported dm-crypt instead of offering only ecryptfs as the main ubuntu installer did for a long time Oct 22 10:29:34 broonie: :) Oct 22 10:30:23 bluelightning: It's unfortunate, but I have to play with the cards I've been dealt Oct 22 10:30:59 you're lucky we've already figured out that ecryptfs has problems with long/deep paths Oct 22 10:31:39 bluelightning: Debian always went with block device encryption AFAIR. Oct 22 10:31:53 Though I guess it's an interesting project for someone to try to fix ecryptfs. Oct 22 10:32:19 broonie: right, the debian installer was sensible (which is what they use on the ubuntu alternate install disk I think, hence they support it there) Oct 22 10:32:53 broonie: possibly, but to be honest the model it uses just seems broken to me Oct 22 10:50:05 woglinde: using http://paste.debian.net/128014/ in my local.conf I get the error you can see in http://paste.debian.net/128108/ (with the patch that works around it). Does the patch make sense? Oct 22 11:26:43 mario-goulart okay maybe I made the mistake or the out of tree builds where changed serval times Oct 22 11:27:05 when I have the time and see it on my buildserver I will fix it Oct 22 12:10:06 bluelightning, if I make changes to a .bbappend file, lets say X.bbappend and then rebuild an image which indirectly depends on X, would that not trigger a rebuild of X and also the image? Oct 22 12:11:00 (it's not in my case, I'm wondering what I'm doing wrong..) Oct 22 12:15:38 I'd say it should (?) Oct 22 12:16:51 my specific case is: X.bbappend adds a few patches to be applied during build. If I remove one of the patches by editing X.bbappend, and rebuild, i don't see bitbake rebuilding X from scratch. I would expect that it would unpack, patch (with all except the patch I just removed) and compile Oct 22 12:25:52 if I do "bitbake -f -c unpack X" and then "bitbake X", it will rebuild without the patch as expected. Im just puzzled why Bitbake doesn't detect my change to X.bbappend automatically Oct 22 12:50:10 mago_: it should yes Oct 22 12:50:18 mago_: what change are you making specifically? Oct 22 12:55:13 hmm I run qmake on the target and get Oct 22 12:55:20 root@ettus-e300:~/RTIMULib/RTIMULibDemo# qmake Oct 22 12:55:21 sh: -d: command not found Oct 22 12:55:41 qmake is a binary Oct 22 12:56:35 bluelightning, i modified SRC_URI to not include a patch anymore Oct 22 12:56:54 mago_: changing SRC_URI should absolutely trigger do_fetch and beyond Oct 22 12:57:12 let me try again Oct 22 12:57:32 * Crofton|work straces Oct 22 13:01:51 . /usr/share/qt4/environment-setup Oct 22 13:01:52 :) Oct 22 13:03:59 bluelightning, i have a git repo for my meta-layer. X.bbappend is in this layer. After I've built the image that depends on X, i swap to another branch of my layer which contains changes to X.bbappend. If I rebuild again, Bitbake does not detect changes and do not rebuild X. How does bitbake detect changed files (to re-parse)? Cause if it checks just on modification date, i think the other branch has an older modification date than the one used in Oct 22 13:03:59 initial build Oct 22 13:05:32 AIUI git changes the modification date to the latest every time it changes a file, even when going back to an older version Oct 22 13:05:42 at least that's what I thought it did... Oct 22 13:11:58 let me check Oct 22 13:14:13 you seem to be right Oct 22 14:49:33 koen, how does ångström generate its feeds? i guess you do more than just building ipks and then "bitbake package-index"? looking at angstrom-feed-configs.bb, it seems you do not split packages over arch, but rather debug, perl, python, gstreamer etc and then you only have 1 MACHINE_ARCH, where as distro-feed-configs looks for all these EXTRA_ARCHS Oct 22 16:12:06 mago_: upload-packages.bb + contrib/sort.sh Oct 22 16:13:42 koen, ok, cool. I had a look at distro-feed-configs.bb btw, would you say the issue here is that we can't tell for sure what PACKAGE_EXTRA_ARCHS we have generated ipks for until the entire build is done, and at this point, it's too late to generate opkg config files? Oct 22 16:15:06 that's one Oct 22 16:15:35 but including all the possible archs leads to more info being on the device (diskspace in /var/lib) and more memory consumption Oct 22 16:15:54 opkg will OOM on 128MB devices Oct 22 16:17:57 oh, i see. I just made a hack-version of distro-feed-configs.bb now where I just ship ${MACHINE_ARCH} and ${DISTRO_FEED_EXTRA_ARCHS}, which is set by each machine config. Do you see any issues with that? Oct 22 16:20:15 (so basically, I replaced PACKAGE_EXTRA_ARCHS with my own variable which is set by the machine config) Oct 22 16:24:52 look at what angstrom-feed-configs does Oct 22 16:25:00 it's the best people have come up with so far Oct 22 17:42:56 hi, all! can I implement image feature, which will build different kernel recipe for particular image? or how this is generally done? Oct 22 17:44:36 is there an easy way to print a single bitbake variable? (i need to access it from a shell script) Oct 22 17:45:26 slapin, i think you can just depend on as many kernel recipes as you want in your image, yes Oct 22 17:45:40 mago_: can sort of be done with bitbake -e and grep, if your grep pattern is careful Oct 22 17:45:55 mago_: otherwise there's a separate tool that provides that capability, it's unofficial -- https://github.com/kergoth/bb Oct 22 17:46:20 from shell, I usually do: eval $(bitbake -e | grep '^MACHINE=') Oct 22 17:46:22 or similar Oct 22 17:46:24 For simple things I usually resort to bitbake -e | grep ^VARIABLE= Oct 22 17:46:28 indeed Oct 22 17:46:38 and how would you extract the actual value of the variable in that case? Oct 22 17:46:42 (not a shell guru) Oct 22 17:46:43 bb show MACHINE or bb show -r MACHINE to avoid the grep, still outputs in var=value syntax Oct 22 17:46:49 mago_: see the eval i just mentioned Oct 22 17:47:00 there are other options with sed, of course Oct 22 17:49:06 mago_: something like echo FOO=val | sed -r 's/[^=]+=(.*)/\1/' Oct 22 17:50:08 or bitbake -e | sed -n -e 's/^\([^=]*\)=['"]\(.*\)['"]/\2/p' to avoid the grep Oct 22 17:50:17 that's nice. The eval approach works for me Oct 22 17:50:20 thanks Oct 22 17:50:35 (the sed i just posted is untested, off the to pof my head, fair warning) Oct 22 17:50:35 kergoth, any plans to merge some of the bb features into bitbake? or are they trying to keep the core clean? Oct 22 17:50:49 yes, we want to move in that direction Oct 22 17:51:31 mago_: but I want one at time, and so, that virtual/kernel is the selected one. is it possible within image scope, or should be single set per TMPDIR? Oct 22 17:53:02 mago_: I have one mainline kernel and another custom patched old vendor kernel, and want to make it selectable. The best if that could be made workable within single TMPDIR, but so, that MACHINE name stays the same... is it possible? Oct 22 17:55:04 slapin, so you want to generate two images for the same MACHINE, with different kernels? Oct 22 17:56:14 mago_: basically yes Oct 22 17:57:31 well, not sure if you can select a completely different kernel using IMAGE_FEATURES tbh. I think the idea is that an image feature changes the configuration, not the actual dependency tree Oct 22 21:19:49 can someone remind me how to check my bbappend is appended? Oct 22 21:26:08 bitbake-layers show-appends Oct 22 21:33:45 thanks kergoth Oct 22 21:33:50 np Oct 22 21:34:15 I am an idiot Oct 22 21:34:24 forgot version in bbappend :) Oct 22 21:34:26 thanks Oct 22 21:34:42 now I need to remember this when I get back from my cultural experience **** ENDING LOGGING AT Thu Oct 23 03:00:00 2014