**** BEGIN LOGGING AT Fri Jun 07 02:59:57 2019 Jun 07 08:50:15 Why do we print a warning message when packaging a SW I have no dependency on is incompatible in terms of licensing? Jun 07 08:50:36 e.g.: WARNING: cairo-1.14.12-r0 do_package: QA Issue: cairo-perf-utils has an incompatible license. Excluding from packaging. [incompatible-license] Jun 07 08:51:06 I do depend on cairo, but cairo, nowhere cairo-perf-utils, why would I care if cairo-perf-utils is not packaged? Jun 07 08:51:21 because bitbake doesn't know that you're never going to use cairo-perf-utils Jun 07 08:51:38 if you were doing a 'bitbake cairo' how would it know that you're never going to use the package? Jun 07 08:55:10 rburton: well, if I'd like to use the cairo-perf-utils package, I'd do `bitbake cairo-perf-utils`, not `bitbake cairo` right? Jun 07 08:55:36 no you wouldn't because you bitbake a recipe and cairo-perf-utils is a package generated by that package Jun 07 08:56:06 the alternative is that bitbake silently makes packages disappear and you have no idea why the recipe says a package exists but it wasn't built. i think i prefer the note. Jun 07 08:56:17 argh, "generated by that recipe" Jun 07 08:56:20 too early Jun 07 08:56:58 ah true, `bitbake recipe` not `bitbake package` Jun 07 08:57:03 :/ Jun 07 08:57:48 rburton: my concern is that Yocto is pretty verbose already Jun 07 08:58:00 silence the warning if you don't care Jun 07 08:58:23 and I'd like our users to actually care about the warning messages and tell us when they encounter one Jun 07 08:58:40 WARN_QA_remove = "incompatible-license" Jun 07 08:58:53 because currently they don't even look at it so it's counter-productive Jun 07 08:58:59 the alternative is to bump most of the warnings to errors so you have to look at them Jun 07 08:59:02 (but not this one) Jun 07 08:59:27 note that poky makes more warnings fatal compared to bare oe-core for this reason Jun 07 08:59:37 interesting Jun 07 09:00:03 WARN_TO_ERROR_QA = "already-stripped compile-host-path install-host-path \ Jun 07 09:00:03 installed-vs-shipped ldflags pn-overrides rpaths staticdev \ Jun 07 09:00:04 unknown-configure-option useless-rpaths" Jun 07 09:00:04 WARN_QA_remove = "${WARN_TO_ERROR_QA}" Jun 07 09:00:06 ERROR_QA_append = " ${WARN_TO_ERROR_QA}" Jun 07 09:00:11 says poky.conf Jun 07 09:00:34 used to be more but i recently pushed the more useful ones to error in oe-core by default Jun 07 09:00:41 this is new :) Jun 07 09:00:54 We were hit by those changes pretty hard :D Jun 07 09:01:09 But I'm happy to have had to fix them Jun 07 09:02:10 I see I also could do an INSANE_SKIP_${PN}-perf-utils += "incompatible-license" Jun 07 09:02:22 is there any difference at all? any better than the other? Jun 07 09:04:30 why would you care about the license warning in cairo and not other packages? Jun 07 09:04:45 if you don't care about that warning, make it go away entirely by removing from WARN_QA Jun 07 09:05:29 I try to avoid silencing on the global level Jun 07 09:06:13 for that license there's an argument that's the right move: if you don't care about the warning, shut it up Jun 07 09:07:33 rburton: thanks a ton for your clear explanation Jun 07 09:13:38 also for anyone wondering, it seems that if there is a QA issue detected by Yocto, if it's neither in WARN_QA nor in ERROR_QA, a bb.note is still raised (helpful in logs or when debugging), so it's not entirely silencing it, just making it discreet, which is an important piece of information for me at least Jun 07 09:17:49 hello folks Jun 07 09:18:26 my yocto-created linux system boots when I'm using the .hddimg, which boots the linux as a live system, the rootfs will be put into RAM Jun 07 09:19:15 but when I boot it using the .wic (genericx86-derived machine with corei7-32 tune), then the kernel can't find the root filesystem. Jun 07 09:20:01 and this only happens when I'm trying to boot from an USB-stick. when booting from an HDD, it works fine. I tried this with real machines and QEMU Jun 07 09:22:23 so, the kernel detects my USB-stick, the usb-storage driver outputs its "USB Mass Storage device detected" message. But I see *no* "sd: ..." lines, as it happens on my desktop system for the "/dev/sda" device to appear Jun 07 09:22:57 litb: rootdelay or rootwait in the kernel command line Jun 07 09:23:10 (you need to add one of them) Jun 07 09:25:13 (rootdelay is the timeout in seconds before giving up mounting the rootfs, rootwait (no argument) is to say never timeout when trying to mount the rootfs) Jun 07 09:27:54 ah, thanks! Jun 07 09:28:11 you're lucky, I had the same issue beginning of this week :D Jun 07 09:28:18 xD Jun 07 13:03:34 Anyone else seeing qemu hang indefenitly as part of PGO in python do_compile? Jun 07 13:07:03 JPEW: no... Jun 07 13:08:11 It's new since I rebased on master yesterday. I'll try to bisect it Jun 07 13:11:34 is there a more lightweight way to have a layer-index, without installing the dejango webapp? Jun 07 13:11:51 i.e something like just a plain folder with subfolders for each layer or something? Jun 07 13:14:43 litb: I think you might misunderstand what the layer index is. It is a web app which shows a collection of layers in a web interface Jun 07 13:15:51 RP, ah i see. Because "bitbake-layers layerindex-fetch" can be used to fetch a layer including its dependencies, I thought that a layer-index is a somewhat more lower-level thing Jun 07 13:16:11 like a structured repository/database of layers of a notion of dependency, or something Jun 07 13:17:00 litb: it is that but layers do have dependency information in them Jun 07 13:17:22 litb: the harder part is that a layer wouldn't list where those dependencies come from which is where the index can help Jun 07 13:50:36 it appears to me that _append and _prepend are subtly different from += and =+ in that _append and _prepend are evaluated after parsing all files by bitbake, but += and =+ are evaluated right at that point during parse? Jun 07 13:51:15 I've nowhere read about that subtlety, but sometimes, I read notes like "Because of internals of bitbake, here you can only use _append, not +=" and the like. Jun 07 13:53:27 litb: https://www.yoctoproject.org/docs/latest/bitbake-user-manual/bitbake-user-manual.html#appending-and-prepending Jun 07 13:55:20 ahh I see! My book "Embedded Linux Systems using the Yocto Project" doesn't mention this at all! Jun 07 14:00:02 litb: have a look at the bitbake manual Jun 07 14:06:59 really cool. Next I guess I'll need to make the fstab use UUIDs instead of /dev/sdX, to make booting from USB-stick reliable. sounds like bbappending base-files is in order Jun 07 14:10:22 RP: I didn't see any patches for the bitbake "multiconfig" rename Jun 07 14:11:23 (on the ML that is) Jun 07 15:46:33 JPEW: I was testing then was going to post Jun 07 15:47:04 OK Jun 07 15:48:23 JPEW: sent now Jun 07 15:52:02 * armpit should be mc-hammer... pictures RP in those pants dancing Jun 07 15:52:27 armpit: with the UK<->US translation of pants this isn't good ;-) Jun 07 15:53:02 trousers better Jun 07 15:55:23 * LetoThe2nd blasts https://www.youtube.com/watch?v=tFelfYq3zpQ Jun 07 16:43:18 RP: I tried to upgrade gettext, it's a mess :( We should probably wait until desktop distros do it, and take their patches Jun 07 16:43:56 what gettext upstream did, they removed many of the m4 macros (because 'no one would need to use them' I guess), and this breaks other components Jun 07 16:47:06 kanavin_: hmm, I guess we could cheat and just patch them back in for now Jun 07 16:50:36 kanavin_: date/time issue was well spotted btw, I wondered what the common denominator was between those failures Jun 07 16:58:22 RP: nice weekend to you, going for dinner :) Jun 07 16:59:46 kanavin_: you too! (and me too!) Jun 07 18:16:52 New news from stackoverflow: Yocto - Create and populate a separate /home partition Jun 07 18:20:34 Odd - that was asked, and answered, on May 17. The last comment added was also on May 17. Why is that just now showing up in this chatroom? **** BEGIN LOGGING AT Fri Jun 07 19:32:58 2019 Jun 07 20:11:36 hm, I'm a bit lost with how popuplate_sysroot is supposed to work Jun 07 20:14:25 with recipe at https://pastebin.com/2kT1zpMN, "bitbake depot-tools-native -c populate_sysroot" produces expected contents in tmp/work/x86_64-linux/depot-tools-native/git-r0/sysroot-destdir/usr/share/ but the sysroot-providers/ dir is really in tmp/work/x86_64-linux/depot-tools-native/git-r0/sysroot-destdir/$PWD/tmp/work/x86_64-linux/depot-tools-native/git-r0/recipe-sysroot-native/sysroot-providers/depot-tools-native Jun 07 20:15:52 as if the sysroot-destdir path was prepended to expected path, but it's not like if I had messed with those myself :) Jun 07 20:27:36 that's expected, afaik. native.bbclass sets prefix to a full absolute path in tmpdir Jun 07 20:27:50 so any paths taht include that will be deeply nested Jun 07 21:51:01 I'm working on moving scons (python-scons python-scons-native scons.bbclass) to python3. any reason to keep the python2 version around? Jun 07 21:51:17 or should I just drop the python2 version. Jun 07 22:28:12 drop it Jun 07 22:28:55 if anyone want the py2 version we can create meta-python2 on OE for that Jun 07 22:32:36 something like meta-glv2 Jun 07 22:32:52 * armpit hmm?? gplv2 Jun 07 23:55:26 armpit: can you comment on the patch series on the ML or on the bug https://bugzilla.yoctoproject.org/show_bug.cgi?id=13381 Jun 07 23:55:27 Bug 13381: normal, Undecided, 2.8 M2, timothy.t.orling, IN PROGRESS REVIEW , SCons should migrate to Python 3 Jun 07 23:55:29 thanks! Jun 07 23:56:16 (BTW, it installs /usr/bin/scons so it will conflict if both py2 and py3 are installed. even more reason to drop it) **** ENDING LOGGING AT Sat Jun 08 02:59:57 2019