**** BEGIN LOGGING AT Mon Oct 31 03:00:01 2016 Oct 31 12:06:43 Hi Oct 31 12:07:21 what it does exactly bitbake image -c populate_sdk ? Oct 31 12:07:28 build image + sdk or only the sdk? Oct 31 12:39:13 is it safe to delete the content from deploy/images? Oct 31 12:39:49 I did bitbake -c clean but the images still there Oct 31 12:45:39 aV_V it's best to only do rm deploy/images/core-image-minimal* or whatever your image is called Oct 31 12:46:09 but that works well Oct 31 12:46:14 ernstp: yeah, I did that Oct 31 12:47:31 btw, I'm rebuilding now and on my local.conf I have DISTRO_FEATURES_remove = "alsa" but I'm seeing tasks performed on pulseaudio, alsa,... packages Oct 31 12:48:11 aV_V, just remember some recipes (other than images) deploy to images/ directly and may not re-deploy on a consecutive rerun of the build. Oct 31 12:59:17 aV_V there's also something called machine features... Oct 31 14:08:59 I have this error Oct 31 14:09:02 dd: failed to open '/home/anatoli/poky/build/tmp/deploy/images/cm-fx6-evk/u-boot-cm-fx6-evk.imx': No such file or directory Oct 31 14:12:37 aV_V: did you delete/move that file in the deploy/images/ directory? (you mentioned you were playing in there before ;) ) Oct 31 14:12:58 yes Oct 31 14:13:03 hmmm Oct 31 14:13:21 I need to do bitbake -c clean u-boot ? Oct 31 14:14:29 aV_V: bitbake -c cleansstate u-boot, is more likely required to re-trigger the deploy task Oct 31 14:17:14 nrossi: nice ty! Oct 31 14:18:32 now fails another thing xD Oct 31 14:18:48 I need to clear all Oct 31 15:03:02 rburton: Hi, I am literally building on debian ;-) Oct 31 15:03:06 rburton: testing Oct 31 15:05:45 Marex: nobody but the world ab runs actually builds u-boot-mkimage for target :) Oct 31 15:07:10 rburton: I do bitbake universe ;-) Oct 31 15:14:37 hi, This is aditya and I am trying to integrate the Wilink8 module with Zedboard and i was following the instruction available at http://zedboard.org/product/wilink-8-adaptor Oct 31 15:15:07 but this is for microzed Oct 31 15:15:45 but i want to work on zedboard Oct 31 15:16:03 i tried but i cud not Oct 31 15:16:35 as i did not get the verification mail Oct 31 15:18:20 argh now slang Oct 31 15:19:28 ARGH Oct 31 15:19:47 someone else who thinks that the moment they release it's sensible to move the old release to /old/ Oct 31 15:19:57 ugh Oct 31 15:20:03 hm Oct 31 15:20:08 i wonder if appending to MIRRORS would be sensible Oct 31 15:20:14 extend it with the /old/ URL Oct 31 15:20:17 if they're going to do that, at least call it 'archive' and put *everything* there, including the new Oct 31 15:20:22 yeah exactly Oct 31 15:20:24 yeah, i've done that before. local mirrors change in the recipe Oct 31 15:20:56 but i think checkuri disables mirrors anyway Oct 31 15:25:39 and now i discover that i don't know how mirrors works ;) Oct 31 15:27:29 the most common point of confusion with mirrors is usually the fact that the regexes are on a per-component basis. first url split, then per-component regex replacement, then re-join, with components that are left out entirely being left alone Oct 31 15:27:33 * kergoth gets more caffeine Oct 31 15:31:08 so checkstatus has never supported fallback to MIRRORS Oct 31 15:31:13 but is that what we want, as that isn't the point Oct 31 15:31:15 hmhmhmhmhmhm Oct 31 15:31:58 kergoth: foes PREMIRRORS_prepend = "http://www.jedsoft.org/release/slang/.* http://www.jedsoft.org/releases/slang/old/ " look right? Oct 31 15:32:02 doesn't appear to work Oct 31 15:32:09 likely doing the expressions wrong Oct 31 15:32:50 is /release/ vs /releases/ correct? Oct 31 15:32:54 seems like a mismatch there Oct 31 15:33:25 you passed my first test, well done Oct 31 15:33:29 hehe Oct 31 15:33:45 bah still doesn't work Oct 31 15:34:21 missing \n at the end, if the var isn't empty, it won't work Oct 31 15:34:26 \n is the separaotr between entries Oct 31 15:34:59 you also probably want to append rather than prepend, otherwise it'll contac tupstream instead of the mirrors the distro or user wants to be using in preference to upstream Oct 31 15:35:04 * kergoth shrugs Oct 31 15:39:57 PREMIRRORS_append = "\n http://www.jedsoft.org/releases/slang/.* http://www.jedsoft.org/releases/slang/old/ \n" Oct 31 15:39:57 works Oct 31 15:40:23 seems silly requiring a trailing newline Oct 31 15:40:37 do anybody know how to integrate the WILINK8 module with Zedboard using Architeck SDK Oct 31 15:41:23 rburton: pretty sure it's not required, but you were prepending without a separator, so an existing entry and your entry would be considered one entry. Oct 31 15:41:41 like missing your space in a space separated list Oct 31 15:42:47 yeah maybe i already had a value set Oct 31 15:42:48 Actually i followed the scripts available at http://zedboard.org/product/wilink-8-adaptor which is for microzed board and i carefully changed this to work with zedboard and linux is also booting on this board but the device is not getting detected Oct 31 15:42:55 oh god this slang recipe is ugly as sin Oct 31 15:46:49 heh Oct 31 16:02:08 of course slang checks for X headers for fun Oct 31 17:37:05 Does devtool modify only work for compiling modified C sources? Oct 31 17:37:26 I'm trying to modify the makefiles in the recipe. but devtool build seems to be using older makefiles. Oct 31 17:37:37 what I do -> Oct 31 17:37:43 devtool modify -x guile Oct 31 17:38:09 nano workspace/sources/guile/module/system/base/target.scm Oct 31 17:38:14 devtool build guile Oct 31 17:38:35 changing a simple print in target.scm doesn't seem to be reflecting in the error Oct 31 17:38:36 :S Oct 31 17:38:50 when you use devtool modify, it changes the source tree to the external one. that's what it's using. itd oesnt' pull the amkefiles from somewhere else Oct 31 17:39:33 most likely you need to clean (bitbake -c clean guile) and rebuild it, as the underlying buildsystem isn't regenerating files if necessary Oct 31 17:40:34 The following does show Oct 31 17:40:34 | NOTE: guile: compiling from external source tree /home/zkakakhel/yocto/poky/build/workspace/sources/guile Oct 31 17:40:42 but yea. looks like i need to clean everything and see Oct 31 17:40:58 have a feeling do_configure gets checked from cache or something Oct 31 17:41:04 thanks Oct 31 18:29:09 10496 (pseudo appending spurious bytes, which should have been NULs but apparently sometimes weren't, to extended attributes) now seems to be fixed in upstream. :) Oct 31 19:13:30 seebs: how would you estimate adding gnu/kfreebsd and gnu/hurd support? given it's (I hope) the same libc, it might be easier than adding support for a whole new OS… Oct 31 19:13:47 Dunno. Oct 31 19:14:09 OS X support took about a week, but it's a bit busted right now and will have issues because of the tighter security in newer OS X. Oct 31 19:14:24 Mostly up to what their dynamic linker does. Oct 31 19:14:30 they aren't release architectures in Debian, but I don't like seeing red boxes in the build stats :) Oct 31 19:14:54 I should probably have a look at fakeroot Oct 31 19:15:00 I don't have any boxes running those, and my supplies of Easily Obtained Hardware are vmware machines and I don't think those are on the supported OS list. Oct 31 19:15:03 as I guess the differences must be mostly the same Oct 31 19:15:10 Not really. Oct 31 19:15:14 fakeroot's MUCH simpler. Oct 31 19:15:59 still that may give me some starting point at least :) Oct 31 19:17:10 it'd be fun to try to rebuild Debian using pseude instead of fakeroot and see what breaks (if anything does at all) Oct 31 19:17:15 pseudo* Oct 31 19:18:33 The biggest difference is the path stuff. Pseudo tries to track paths, fakeroot doesn't. Oct 31 19:19:02 This was a problem about two build systems back for WR because build trees had a single fakeroot database which was supposed to be persistent for longish periods of time. Oct 31 19:19:16 Also it was a problem because, if inode reuse got you bogus results, there was nothing you could do to isolate it really. Oct 31 19:19:46 I note, BTW, some of the performance improvements have reduced reliability, and I have plans to fix that at some point. Oct 31 20:08:05 How can I run a build in devshell? Oct 31 20:08:23 If I just "make", I'm missing include files from another library, but the recipe works when run directly. Oct 31 20:09:42 you need to run the do_compile task, not make. run compile at least once with bitbake (just bitbake recipe or bitbake -c compile recipe), then in devshell, run ../temp/run.do_compile Oct 31 20:12:40 I guess I expected all the export lines at the top of run.do_compile to have been run in the devshell. Oct 31 20:20:35 Circuitsoft: and? the exported vars are minor at best. it's the actual build commands from the do_compile task you care about, and we can't inject shell functions into the spawned shell easily Oct 31 20:21:12 Circuitsoft: just exporting vars doesn't necessarily mean they'll be used in the underlying buildsystem, oftentimes we pass the on the make commandline. and oftentimes the build involves more than just running a single 'make' Oct 31 20:21:22 read the recipe for details Oct 31 20:27:45 kergoth: Well, in this case, it is the exports that make things work. The actual build command comes down to just "make". Oct 31 22:39:46 How do I get a native toolchain on my device. I tried adding gcc/automake to IMAGE_INSTALL_append Oct 31 22:41:34 lolsborn: you may find packagegroup-core-buildessential useful Oct 31 22:44:32 That's what I was looking for. Thanks. Nov 01 02:53:43 Hi all, I'm building an image using wic and a .wks file. I have one extra partition which is empty, but I need to change the owner uid of the root directory of that partition to somethign else than root (uid 0). How can I achieve this? I can't find anything... Nov 01 02:56:53 Basically, I would need to do the equivalent of 'losetup && mount /dev/loop/p5 partition5 && chown 1000:1000 partition5/ && umount partition5' Nov 01 02:57:01 but through wic. Nov 01 02:57:08 Is that possible? **** ENDING LOGGING AT Tue Nov 01 03:00:01 2016