**** BEGIN LOGGING AT Tue Jun 01 02:59:57 2021 Jun 01 07:33:49 good morning Jun 01 07:51:34 hi, is it possible to use different version of a native tool to build different packages? I need a tool to generate code for 2 different packages but I need to use 2 different version of the tool. I don't need it to be installed in the target image. I tried to add (= version) in the DEPENDS variables in the recipes but bitbake ignores it Jun 01 08:02:36 Chep: easiest method is probably having two recipes and naming them differently, foo1 and foo2 and put those on the according DEPENDS Jun 01 08:02:43 mckoan: morning Jun 01 08:03:29 alejandrohs: thanks Jun 01 08:03:50 halstead: I'm getting name resolution errors from git.yp, probably nothing major, just fyi in case you see them again Jun 01 08:05:34 is docs.yoctoproject.org down? 😥 Jun 01 08:11:06 It's like the domain doesn't point to any server Jun 01 08:11:22 but I'm not sure if this is because my DNS 🤔 Jun 01 08:31:55 wyre: I guess both issues are related Jun 01 08:42:42 alejandrohs, both issues? what two issues? Jun 01 08:42:59 I just had one, but now apparently is up again docs.yoctoproject.org Jun 01 09:52:52 tlwoerner: one thing I'm not comfortable with in my Rock64 support patch is that need to tweek RK_BOOT_DEVICE to generate a sd or emmc image ending up with the same filename. Maybe the best course of actions would be to make device-independant images that could be flashed on both media types, discovering where they are booted from Jun 01 09:57:41 not sure how much this kind of things could be done without having to touch wic Jun 01 10:01:17 what is the "--ondisk ${RK_BOOT_DEVICE}" good for anyway ? I'd tend to think there should be no need to know if we're going to use mmcblk0 or mmcblk1 to create the partition table. Then we'd only need to get the root= parameter generated from the u-boot bootcmd, rather than being hardcoded in the wks, right ? Jun 01 10:06:02 an alternative could be to provide core-image-*-{sdcard,emmc} variants, but I feel that would not feel greatly integrated with the rest Jun 01 10:48:02 PSA: I'm moving to libera channel only, see you there Jun 01 15:12:50 Does changing ERROR_QA in a machine config change ERROR_QA for all packages pulled in? Or just the final machine image Jun 01 15:14:21 Does changing ERROR_QA in a machine config change ERROR_QA for all packages pulled in? Or just the final machine image Jun 01 16:37:42 * paulg wonders why image-mklibs class got voted off the island. Jun 01 16:39:06 Not that I care or know what it did ; just that since it was a part of our boilerplate conf/* content, nuking it broke every existing build dir out there with a meaningless screaming red ERROR parse. Jun 01 16:40:13 ...and to be clear, I say "our" meaning "Yocto" and not any specific company variant of Yocto. Jun 01 16:40:40 paulg: I noticed that problem here too :) Jun 01 16:41:04 paulg: look into what mklibs does, then wonder why it didn't exit the stage much earlier ;-) Jun 01 16:41:54 thinking dummy-ing out the file and replacing it with a bbwarn telling people to edit their conf line such-and-such would have been more end user friendly. Jun 01 16:43:00 I often wonder if zeddii and I are outliers who re-use the same build dir over and over and over and over .... Jun 01 16:43:28 everyone else does a nuke-and-pave and relies on sstate to save them? Jun 01 16:43:40 Dunno. Jun 01 16:44:38 sstate could vanish overnight and I'd probably not notice, or perhaps even be happier for it. Jun 01 16:45:34 kinda like distcc and ccache type stuff. Seem like dated solutions to a 2005-ish problem. Jun 01 16:49:05 at least they aren't like autoconf - a solution to a 1985 problem. B-) Jun 01 16:49:47 Guess it managed to out-live alt.autoconf.die.die.die and NNTP in general. Jun 01 16:50:16 paulg: sstate rocks and the autobuiler would grind to a halt without it. We could have added dummy mklibs files but it is just changing a line in a file ;-) Jun 01 16:50:52 but the autobuilder is not indicative of the typical end user. Jun 01 16:52:52 paulg: I do use it heavily locally too and it is useful to users when they clear tmpdir out in totality Jun 01 16:53:54 Ah yes, my favourite error message. The one that comes with a TL;DR saying you need to nuke all of tmp because something unexpected happened. Jun 01 16:54:18 i wipe tmp with nearly every build, it's harmless and restoring from sstate is fast. i wouldn't want to build without it Jun 01 16:54:21 paulg: I think that one should be a lot less common now... Jun 01 16:55:18 kergoth, and eat the ensuing i/o storm that comes with it? Why? Jun 01 16:56:48 paulg: I'm not even going to attempt to convince you after zeddii's recent comments about kernel devs and workflows ;-) Jun 01 16:58:08 Probably wise. Don't argue with the old fart who has "KEEP OF MY LAWN" signs. Jun 01 16:58:46 Save it for when I send my patch series to remove sstate. Jun 01 17:03:09 how is a kernel specified within an image recipe? e.g., if i am extending core-image-minimal, how do i specify which kernel it will use? Jun 01 17:04:18 i've googled around, looked at several yocto docs, don't see it. Jun 01 17:26:26 yates: Typically you specify that using the PREFERRED_PROVIDER_virtual/kernel variable. Jun 01 18:56:09 hello, Jun 01 18:58:08 I have added IMAGE_INSTALL_append = " mariadb" to the local.conf but most apparently the mariadb server does not get installed (no my.cnf in /etc/) tried mariadb-server but nothing provides that, how do I make sure  the server gets instaled in the image? Jun 01 20:52:37 Saur: thank you. Jun 02 00:22:10 have folks moved over to the new irc server (libera chat)? Jun 02 00:36:06 many have **** ENDING LOGGING AT Wed Jun 02 02:59:57 2021