**** BEGIN LOGGING AT Wed Sep 26 03:00:01 2018 Sep 26 09:07:00 Hello all! Sep 26 09:08:46 Got a questuin. In one of the recipes I sign the image and my private key is protected by a passphrase. During the bake I've got a error: Enter pass phrase for privkey.priv User interface error unable to load key. The reason is obvious: openssl is ran detached from console. Is there a way how to solve this? Sep 26 09:20:39 see the existing gpg signing code which lets you provide a source for the passphrase Sep 26 09:24:01 rburton, could you please give me a hint what is the recipe for this? Sep 26 09:25:06 gpg_sign.py is where the gpg signing happens. the key is that you give the passphrase directly to the tool instead of expecting the user to sit there and wait for the passphrase to be entered Sep 26 09:25:20 use a agent, or have it waiting on a pipe, or just put it in the local.conf as a string Sep 26 09:27:45 rburton, that's what I wanted to avoid... Will use env variable then. Thank you!! Sep 26 09:51:11 Kitsok: env variables will be stripped by default. safer to use an agent as they're designed for security Sep 26 11:18:55 rbutton, you mean SSH agent? Sep 26 11:19:00 hi all, question, can yocto run on a debian sid ? Sep 26 11:19:45 angelo_ts: run yocto? what do you mean? Sep 26 11:21:01 LetoThe2nd, mm i mean if it can build on a debian sid host. Debian "sid" seems not in the supported distributions Sep 26 11:21:24 angelo_ts: i build on debian testing (buster), but in a chroot Sep 26 11:21:26 angelo_ts: it *might* work, and if it breaks then fix it and send patches :) Sep 26 11:24:17 angelo_ts: should work but as sid is a rolling target, it may not. Sep 26 11:24:36 for example if you get a new gcc which can't compile an older gcc, then it breaks Sep 26 11:25:44 ok thanks. Sep 26 11:26:12 in case, what alternative should be better between chroot or a poky docker container ? Sep 26 11:26:28 well, feel free to try sid Sep 26 11:26:44 chroots are trivial though Sep 26 11:30:26 i personally like docker, but hey, thats just me. Sep 26 11:30:41 yeah whatever Sep 26 11:30:49 i've a pile of chroots that i 'boot' using nspawn Sep 26 11:30:52 same difference Sep 26 11:31:23 ok thanks Sep 26 11:38:54 hi, I had a systemd Breaking ordering cycle by deleting job alignment.service/start. to remove this i removed the startop of the /etc/init.d/alignment.sh file. Sep 26 11:40:17 I woold like to know what the consequences are from removing: echo "3" > /proc/cpu/alignment in the boot process on my arm target Sep 26 13:39:45 Hi everyone! Noob question: is there any reason why the task do_kernel_configme could not be available for the target core-image-minimal ? Sep 26 13:39:58 a.k.a ../bb.sh core-image-minimal -c kernel_configme -f Sep 26 13:40:16 results in ERROR: Task do_kernel_configme does not exist for target core-image-minimal Sep 26 13:45:46 because kernel_configme is a kernel task? Sep 26 13:45:51 and core-image-minimal isn't a kernel Sep 26 13:51:02 I thought it would be kernel compiling + rootfs creation + image creation Sep 26 13:51:10 no Sep 26 13:51:10 I should have verified. Thanks! Sep 26 13:51:17 images are constructed from packages Sep 26 13:51:25 all an image recipe does it unpack a load of packages Sep 26 13:59:21 So if I want to have a core-image-minimal with my own .config, I need to call ../bb.sh core-image-minimal after creating the kernel with ../bb.sh linux-yocto? Sep 26 14:27:10 New news from stackoverflow: No Module named 'datetime' in python3 when creating a recipe for Yocto but datetime exists in python2.7 Sep 26 14:50:53 in an image recipe, is there a way to append a flag in IMAGE_FEATURES if a flag is already present in this variable without triggering a RecursionError? My use case is to append read-only-rootfs if debug-tweaks is not set in IMAGE_FEATURES Sep 26 14:51:27 IMAGE_FEATURES_append = " ${@bb.utils.contains("IMAGE_FEATURES"...)}" obviously doesn't work Sep 26 15:04:20 What is the difference between "poky" and "poky-lsb" images? Interested in both design and outcome differences. I have built one, but not another yet. Sure can do, but hope to get some clue faster by asking. Sep 26 15:05:54 I couldn't track it down on my own - what I see is poky-lsb "overrides" "linuxstdbase" or such, doesn't say much to me, and I haven't dug deeper yet. Sep 26 15:26:25 RP: any objections to a bb.data.hasTask method to identify if a task exists? Sep 26 15:26:49 instead of d.getVar("do_foo", "task") Sep 26 15:38:20 rburton: should that be bb.build.hasTask() ? Sep 26 15:38:33 erm yeah tat Sep 26 15:38:34 rburton: the other *Task funtions are in there iirc Sep 26 15:38:45 rburton: no objection though Sep 26 15:39:27 rburton: even if you think my feedback is tat ;-) Sep 26 15:39:33 how rude Sep 26 15:39:50 ha Sep 26 15:39:55 typing failure as usual Sep 26 15:45:44 so archiver.bbclass has a 'configured' mode Sep 26 15:45:49 which is madness but still Sep 26 15:46:11 actually first question is who actually wanted configured mode, and what's the point Sep 26 15:48:56 but other question Sep 26 15:49:12 currently it just hooks do_ar_configured after do_configure, then re-runs do_configure with a custom WORKDIR Sep 26 15:49:29 which then has lots of special casing for evertyhing that injects tasks between unpack and configure Sep 26 15:49:41 so, alternative Sep 26 15:50:20 do_ar_configured depends on nothing but does bb.build.tasksbetween(do_unpack, do_configure) and runs those itself Sep 26 15:51:40 so 1) is there a better way to know the 'first' task than hardcoding do_unpack Sep 26 15:51:54 2) is there a better way to run tasks directly other than iterating through prefuncs/task/postfuncs Sep 26 17:51:37 can I bbappend a bbclass? Sep 26 17:53:50 RP: no Sep 26 17:53:54 erm jdel no Sep 26 17:57:27 barf. Is there any way to fix a broken class from a new layer? Sep 26 18:15:05 jdel: you can append to the bb file that inherits the class, if that helps, but you probably should just fix the class instead Sep 26 18:22:47 here's a small patch, maybe someone can comment on it: Sep 26 18:22:51 --- a/poky/meta/classes/package.bbclass Sep 26 18:22:51 +++ b/poky/meta/classes/package.bbclass Sep 26 18:22:51 @@ -2158,6 +2158,7 @@ addtask package after do_install Sep 26 18:22:51 Sep 26 18:22:51 PACKAGELOCK = "${STAGING_DIR}/package-output.lock" Sep 26 18:22:53 SSTATETASKS += "do_package" Sep 26 18:22:56 +do_package[depends] += "${POPULATESYSROOTDEPS}" Sep 26 18:22:58 do_package[cleandirs] = "${PKGDEST} ${PKGDESTWORK}" Sep 26 18:23:01 do_package[sstate-plaindirs] = "${PKGD} ${PKGDEST} ${PKGDESTWORK}" Sep 26 18:23:03 do_package[sstate-lockfile-shared] = "${PACKAGELOCK}" Sep 26 18:23:17 do_package utilizes objdump to calculate RDEPENDS Sep 26 18:23:29 but it's possible for do_package to run without binutils being populated Sep 26 18:24:04 at least in the crazy qualcom bbclass i've been debugging Sep 26 19:29:48 the plymouth recipe seems to be essentially designed to work with an initrd - that feature is possible to disable, but then it seems there is no integration whatsoever with the rest of the system (as opposed to pslpash), or am I missing something ? Sep 26 19:38:01 oh, nasty circular dependency when systemd is configured with cryptsetup :( Sep 26 19:54:56 jdel: hmm, you're probably right Sep 26 23:53:24 RP: sweeter words have never been spoken **** ENDING LOGGING AT Thu Sep 27 02:59:59 2018