**** BEGIN LOGGING AT Mon Dec 16 03:01:54 2019 **** BEGIN LOGGING AT Mon Dec 16 07:27:10 2019 Dec 16 07:55:02 good morning Dec 16 07:55:55 you too Dec 16 08:10:50 anybody managed to make recipe for scipy? Dec 16 08:12:11 Hey, I'm switching from debian to yocto and I'm stucked at adding openjdk(v11) in yocto. Is it really not supported by yocto? Has somebody some information? Dec 16 08:17:49 New news from stackoverflow: Define new Imagefeatures with yocto Dec 16 08:18:15 nickes: are you generally having understanding problem with the yocto mindset/workflow, or just cant find java recipes? Dec 16 08:18:26 nickes: in the latter case, the best guess ishttp://git.yoctoproject.org/cgit/cgit.cgi/meta-java Dec 16 08:20:44 LetoThe2nd: I'm aware of meta-java. But as far as I know they only have openjdk v7 and v8. Dec 16 08:21:05 nickes: then its probably recipe writing time for you Dec 16 08:21:53 nickes: java is a problem child in the OE world more often than not, sorry to say. Dec 16 08:22:30 LetoThe2nd: I somehow knew it... Thanks Dec 16 08:22:38 nickes: have fun! Dec 16 08:23:48 I want to patch aktualizr (https://github.com/advancedtelematic/meta-updater/tree/master/recipes-sota/aktualizr) and I need to add a dependency to it. I've created an append file which adds the dependency. The issue is that the aktualizr recipe also wants to build for the native machine where the dependency doesn't exits (and I don't any reason to install it). Is there a way to add the dependency Dec 16 08:23:49 only for the target build? Dec 16 08:26:57 rcrudo: is it actually necessary that you have the -native build? Dec 16 08:27:05 rcrudo: if not, get rid of it. Dec 16 08:27:46 rcrudo: if, then my personal guess would be to rip out all common parts into a .inc, and make seperate target and native recipes that pull it in. Dec 16 08:27:52 (but there might be better ways) Dec 16 08:29:09 LetoThe2nd: no, I don't need the native. But the native build is there (upstream) by default. Can disable it via append? Dec 16 08:29:33 rcrudo: good question. rule #1: try and find out. Dec 16 08:54:55 how can I check what version of Python bitbake is trying to use while baking a recipe? Dec 16 08:55:53 I get RuntimeError: Python version >= 3.5 required even though I have 3.5.6 so I guess it is trying to use python2, how can I change that? Dec 16 09:10:11 LetoThe2nd: the problem is that I don't even know even know what to try. The original aktualizr recipe has BBCLASSEXTEND = "native" Dec 16 10:01:28 morning Dec 16 10:02:44 morning Dec 16 10:03:00 JaMa: with that docs patch you mentioned, we should point it out to scott btw Dec 16 10:05:29 RP: is srifenbark@gmail.com still the correct address? He was in Cc Dec 16 10:06:06 JaMa: yes. I couldn't see the cc on list, thanks Dec 16 10:06:09 RP: and I'm not sure why the last "ping again" got duplicated, I'm pretty sure I've sent it only once Dec 16 10:08:17 hmm on the duplicated e-mail (which seems like sent by the ML itself) he isn't in Cc, so not sure what's going on Dec 16 10:48:18 New news from stackoverflow: improve deployment time of yocto based development Dec 16 10:53:58 rcrudo: naybe just BBCLASSEXTEN = "" ? Dec 16 10:58:26 LetoThe2nd: I've tried that. It didn't work. _append_class-target seems to be doing the trick Dec 16 11:00:22 rcrudo: :) Dec 16 11:12:55 marble_visions: to be fair, DISTRO was my first guess Dec 16 11:14:02 as far as i can tell, the effect will be exactly the same, so its a matter of convention Dec 16 11:15:28 i later said machine because i saw that on my set up the deploy path is based on machine, distro isn't even referenced Dec 16 11:16:34 so it looked to me that if you wanted a build at the end of which you end up with two deployments (rel/dev) the most convenient way to do that will be in machine conf Dec 16 11:20:17 anyone got a build with fvwm to hand? Dec 16 11:46:30 marble_visions: but that might be specific to my setup - maybe distro is a better option, you might ask LetoThe2nd for the opinion, if you care to explain the context to him Dec 16 11:46:41 my context? Dec 16 11:46:43 BEER! Dec 16 12:15:20 LetoThe2nd: :) Dec 16 12:17:49 milloni: :) Dec 16 12:27:45 kanavin: maybe we should split your patchbomb up a little :) Dec 16 12:27:59 rburton, yes :) Dec 16 12:28:24 clearly needs to be merged piece by piece, I am looking at gettext now from RP's mail Dec 16 12:28:55 i've some follow-on pieces for the gettext saga too Dec 16 12:29:00 like switching to autopoint by default Dec 16 12:30:25 kanavin: the attr thing khem reported is what i hit too Dec 16 12:30:50 rburton, yes, I suspect I have a header on my system that isn't present on yours Dec 16 12:30:52 kanavin: maybe you have those headers on your host Dec 16 12:30:56 $ find /usr/include/ -name xattr* Dec 16 12:30:59 no attr/xattr.h Dec 16 12:31:14 sorry half my paste disappeared as it was /usr ... Dec 16 12:31:21 but no attr/xattr. Dec 16 12:31:23 alexander@alexander-box:~/development/poky/meta/recipes-kernel/systemtap$ find /usr/include/ -name xattr* Dec 16 12:31:42 hm Dec 16 12:31:51 usr/include/attr/xattr.h Dec 16 13:18:44 New news from stackoverflow: php://input stream is empty on lighttpd and yocto based image Dec 16 14:06:56 I was getting worried, not seeing #yocto highlighted in irssi for the whole day.. I was not on the chan anymore. What happened? Netsplit? Dec 16 14:08:46 qschulz: I had the same issue, think it was a netsplit Dec 16 14:11:23 RP: first time ever that happened for me :) Being curious now, anyway to reconnect automatically after a netsplit with irssi? Dec 16 14:12:06 qschulz: usually the splits heal, I guess some server hasn't. Hard for the client to tell :/ Dec 16 14:15:24 kanavin: I put a number of your other patches in -next to test. Unfortunately there is some bitbake gremlin in there with hashequiv which skewed the results :( Dec 16 14:15:42 kanavin: retrying now with some bits removed Dec 16 14:28:38 RP, thanks Dec 16 15:49:07 New news from stackoverflow: Linux device tree can't find an element declared right above it Dec 16 16:20:11 what is the "official" git tree for bitbake? is it https://git.yoctoproject.org/cgit/cgit.cgi/poky/tree/ Dec 16 16:21:01 actually, is it http://git.openembedded.org/bitbake/ ? Dec 16 16:21:30 milloni: yes Dec 16 16:21:51 the latter? Dec 16 16:22:08 milloni: yes Dec 16 16:22:14 thanks Dec 16 16:47:29 I'm trying to write a recipe for a cmake application that uses boost. Recipe DEPENDS on "boost" and RDEPENDS on "boost-system", but I'm still getting undefined reference errors (to boost::system stuff). I'm wondering if the "inherit cmake" doesn't pick up the compiler flags? Anyone had this problem? Dec 16 16:51:40 jofr: are these undef refs popping during link stage ? Dec 16 16:53:49 Nevermind. PEBKAC. Dec 16 16:55:14 fwiw RDEPENDS should be redundant, the system will do those for you Dec 16 16:55:20 unless you're dlopen()ing Dec 16 16:57:20 This was just a b0rked CMakeLists.txt that didn't properly add the required libraries. "It works on my machine!" sort of thing. Dec 16 16:57:48 jofr: OK Dec 16 16:58:10 rburton: now that you are at keyboard I see fvwm2.1 -> fvwm.1 is dangling symlink Dec 16 16:58:31 rburton: so your guess is right Dec 16 16:59:33 khem: cool, glad my blind fix was right :) Dec 16 16:59:59 and it also finds a problem in the package which is all good, but it would be good if this class could tell this in a nice humane way instead of poo-pooing over Dec 16 17:00:12 yeah it just skips symilnks now Dec 16 17:00:20 just works on normal files Dec 16 17:00:56 rburton: good Dec 16 17:02:20 rburton: so this is the source of problem http://cgit.openembedded.org/meta-openembedded/tree/meta-oe/recipes-graphics/fvwm/fvwm/0001-Fix-compilation-for-disabled-gnome.patch#n47 Dec 16 17:07:04 https://github.com/fvwmorg/fvwm/commit/0fdf2815bc09edaf4c0ecb45d47ad12d00863239 Dec 16 17:07:10 might fix it Dec 16 17:30:27 hmm, http://www.abisource.com/downloads/enchant/ is down leading to bitbake-selftest failures :( Dec 16 17:32:43 not so enchanted is it Dec 16 17:36:57 maybe we shouldn't rely on ancient software for selftest :/ Dec 16 17:37:07 at least, point at the mirrors Dec 16 17:48:55 rburton: I think it needs the index there but we should just copy it and use from our own Dec 16 17:50:11 is it part of the upgrade testing then? Dec 16 17:50:21 maybe make it do opkg-utils or something on the servers we host Dec 16 17:50:50 1.1.6 of enchant in oe-core I wonder if we should use 2.x Dec 16 17:51:14 1.6.0 I meant Dec 16 17:56:43 perhaps using 1.6.1 with updated uri https://github.com/AbiWord/enchant/releases/download/enchant-1-6-1/enchant-1.6.1.tar.gz might be least intrusive Dec 16 17:57:45 The hash equivalency, is it meant to be use mostly on developer machines and not on CI? Dec 16 18:00:16 roussinm: there is no limitation like that Dec 16 18:01:13 khem: I understand that, but I feel that there is no advantage of activating the feature on CI. Dec 16 18:05:45 roussinm: there is if you dont want to avoid rebuilding dependencies which it now detects should not really be rebuilt Dec 16 18:05:54 so I think CI will benefit too Dec 16 18:37:24 khem: from richard explaination here: http://lists.openembedded.org/pipermail/openembedded-core/2019-August/285300.html, I don't see it for CI, I am probably missing something? Dec 16 18:59:42 roussinm: thats not all. read https://lwn.net/Articles/804640/ section "Better optimization" especially para 4 Dec 16 19:07:03 Hello guys! I have an easy question for you guys. What is the difference between `yocto-layer` and `bitbake-layers` when it comes to create a new layer? Dec 16 19:09:05 I am guessing yocto-layer has been phased out and bitbake-layers took its place? Dec 16 19:12:38 khem: So it seems it will benefit CI too. One question, I guess we should have a global equivalence hash server for the company and get developers connected to it... Dec 16 19:13:55 are Congatec considered a good manufacturer? I'm more of a software guy, not familiar with what's out there. Dec 16 19:14:20 we ordered a SMARC 2 evaluation board from them Dec 16 19:21:17 roussinm: depends upon your setup but yes something on those lines Dec 16 19:22:23 has anyone ever had this output https://paste.centos.org/view/fccab102. It happened to me after adding a layer from bitbake-layers. I am unable to do anything with bitbake-layers Dec 16 19:23:51 Otherwise, where can I report that bug? Dec 16 19:38:51 GrimSleepless: looks very much like you just added a path that is not a layer. Dec 16 19:55:54 LetoThe2nd: That's weird. Cause I git cloned meta-realtime from OpenEmbedded and added it through bitbake-layers. Dec 16 19:57:03 GrimSleepless: have you checked that whatever bitbake-layers put into your conf/bblayers.conf is actually the right path? Dec 16 19:57:41 GrimSleepless: i don't doubt the layer per se, just think that bitbake-layers doesn't check if what you gave it is actually a proper layer. Dec 16 20:00:35 GrimSleepless: plus, especially when it comes to latest master of poky, there have been changes concerning python. so you might be better off sticking to an older release. and even more, the layer doesn't seem to get much love. Dec 16 20:04:49 LetoThe2nd: Thank you for your suggestions. I am currently using Zeus so I guess I should use more like warrior? Dec 16 20:05:53 GrimSleepless: zeus should be unproblematic, as long as you have explicitly set all layers to that branch Dec 16 20:06:40 LetoThe2nd: Yes, that's exactly what I am doing. And the path in BBLAYERS is OK. I will try to figure out what went wrong :S Dec 16 20:07:25 GrimSleepless: any it works without meta-realtime being in there? Dec 16 20:08:20 LetoThe2nd: You were right, there is missing layer. openembedded-layer in my case! Thank you so much for your help and suggestions. Dec 16 20:08:30 GrimSleepless: np, have fun! **** BEGIN LOGGING AT Mon Dec 16 22:40:20 2019 Dec 16 23:04:59 zeddii: congrats on your first build :) Dec 16 23:05:24 :D I see one failure, have to look into that. but yes! Dec 16 23:05:45 oh. oe-selftest, doesn't look related to my changes. some sort of setup issue. Dec 16 23:08:21 RP: one quick "reading the status page question". It looks like qemumips is still running, but when I clicked for details, it seems to be done. Dec 16 23:08:28 Does anyone know the status of the videos recorded at the Yocto Project summit last month? Dec 16 23:09:09 behan accepts bribes. Dec 16 23:09:14 zeddii: its bitbake-selftest failing in the fetcher due to a webserver being down Dec 16 23:09:44 zeddii: qemumips is still going Dec 16 23:10:08 zeddii: still going: https://autobuilder.yoctoproject.org/typhoon/#/builders/60/builds/1353 Dec 16 23:12:06 RP: ok, so I am reading that right! Dec 16 23:12:40 I'll watch it and hopefully be able to send out my kernel -stable series with some extra confidence. I'm glad I put this barrier/learning in my way. Dec 16 23:13:05 zeddii: it would be good to document running builds on the wiki, try and improve the info about this Dec 16 23:13:30 zeddii: it adds some confidence :) Dec 16 23:21:37 zeddii: now its done :) Dec 16 23:29:50 w00t. Dec 16 23:30:22 I need to actually queue one with 4.19 now that I think about it, since that is still LTS. Dec 16 23:31:00 zeddii: you can do interesting things like tweaking the preferred versions for specific testing should you need that Dec 16 23:31:25 yah. that's what I need to figure out. that's through the same 'quick' that I just used ? Dec 16 23:31:35 push a commit that sets it ? for example ? and then that'll kick in ? Dec 16 23:31:49 zeddii: right, just set the config on the test branch Dec 16 23:31:55 cool. will do that. Dec 16 23:32:15 then I'll have the LTS and 5.2 changes tested by tomorrow. Dec 16 23:32:47 and then back to 5.4 qemumips64. the fun keeps on rolling ;) Dec 16 23:32:59 zeddii: I was wondering about 5.4 Dec 16 23:33:18 zeddii: I have my hands full with hashequiv :( Dec 16 23:33:36 some scenarios have really really bad performance :( Dec 16 23:34:03 fray: your changes show up on the profile work I've done, sadly :( Dec 16 23:34:24 eeech. mips64 may become a holiday project at this point, since like everyone, there's other things being squeezed in before I have to head off. Dec 16 23:34:56 but at 5.4 may be the only "versioned" kernel in the the sprint release (with -dev being newer), so I have to sort that out. Dec 16 23:35:03 everything else has been working for over a month now. Dec 16 23:35:26 alt.mips64.die.die.die ;) Dec 16 23:37:57 zeddii: fixing runqueue performance is looking like mine Dec 16 23:38:15 I think we have some nice general runqueue speedups just landed in bitbake master though Dec 16 23:38:44 cool. so some progress. Dec 16 23:49:31 zeddii: some, but a problem remains with some workloads Dec 17 00:48:22 zeddii: do you have ppc64 qemu profile as well generally it should build from tip of tree but wanted to ask **** ENDING LOGGING AT Tue Dec 17 02:59:58 2019