**** BEGIN LOGGING AT Sun Dec 04 02:59:57 2022 Dec 04 08:05:34 Herrie: ah thanks, could be interesting Dec 04 08:42:13 Tofe: I don't use it much, but sometimes it's good for reference and checking logs Dec 04 08:46:48 And I wanted to check the notifications Dec 04 08:47:14 looks like it simply creates a fullscreen semi-transparent window Dec 04 08:48:33 a "toast" is a bit better, but it looks like it's managed by the window manager Dec 04 08:50:11 so it looks like the "dashboard" notification concept isn't there anymore Dec 04 08:55:48 Tofe: Also for the indicators? Dec 04 08:56:11 That looks like something we would use for notifications since it's "floating" Dec 04 08:59:47 I mean the wifi etc ones that were shown on their site, which seem to show as overlay on top right Dec 04 09:15:25 But I guess the challenge would be to render HTML content inside of it? Dec 04 09:15:42 Without needing to have QtWebEngine too Dec 04 09:42:17 JaMa: I seem to get compile errors on webkitgtk even with 32gb swap Dec 04 09:42:57 fatal error: Killed signal terminated program cc1plus Dec 04 09:43:06 Is this due to OOMK or something else? Dec 04 12:20:20 Herrie: well, it's always possible to use webruntime for the rendering; the main challenge is to setup the popup size Dec 04 12:20:56 Herrie: webkitgtk ? what are you trying to build? Dec 04 12:34:57 Tofe: JaMa's oe stress test on laptop Dec 04 13:07:08 Herrie: check dmesg, OOMK is shown there Dec 04 14:33:29 JaMa: OK will do when back home later today Dec 04 17:06:28 JaMa: Indeed seems some OOM there for ld and cc1plus Dec 04 17:06:36 Should I add more swap? Dec 04 17:06:53 I have 32GB mem and 32 SWAP now Dec 04 17:07:16 With the "all" and "8" that seems to lead to OOM still Dec 04 17:48:25 "2" passed already? Dec 04 17:49:08 it will be terrible performance due to swap, but yes adding more should let it pass Dec 04 17:55:57 I don't know oe stress test, but isn't there a better way to fit the actual hardware? I mean, stress testing the swap doesn't make much sense to me :) Dec 04 17:58:13 Also, maybe I could add my own builder to the list? Dec 04 18:02:59 Tofe: https://github.com/shr-project/test-oe-build-time Dec 04 18:03:11 JaMa: already executing ./test.sh :) Dec 04 18:03:21 and yes I would be happy to have your builder on the list Dec 04 18:03:38 I chose hardknott, that's the correct one, right? Dec 04 18:04:27 "fit the actual hardware" yes, there are better ways, but the IMHO-wrong default values in oe-core and OSE are number of threads for both Dec 04 18:05:03 and part of the reason for this script is to show how badly that works for "big" builds on machines without insane amount of memory Dec 04 18:05:32 mmh "ImportError: cannot import name 'Mapping' from 'collections' " maybe I shouldn't be using python3 Dec 04 18:05:57 it won't happen when buiding core-image-minimal (as it doesn't have even the webkit-gtk from oe-core, nor any other memory hungry recipes Dec 04 18:06:45 and for OSE/LuneOS it will happen sometimes (when webruntime/nodejs happen to run in paralel with something a bit hungry) Dec 04 18:07:09 as on 16t laptop it can still spawn 16x16 cc1plus processes Dec 04 18:07:33 oh. That would be a OOM for me too Dec 04 18:07:37 (if unlucky) Dec 04 18:07:43 Tofe: hardknott has most of the results, but I would test kirkstone now Dec 04 18:07:53 Herrie: which release are you testing? Dec 04 18:07:56 ok Dec 04 18:08:05 Tofe: and for Mapping there is simple backport for bitbake Dec 04 18:08:38 https://git.openembedded.org/bitbake-contrib/commit/?h=jansa/1.50&id=2805102e2a142c8e821519b651e4035353bbb717 Dec 04 18:09:09 thx Dec 04 18:09:09 strange you don't have this, let me check the revisions in hardknott branch Dec 04 18:09:33 I'm now on kirkstone Dec 04 18:10:50 hmm hardknott is quite a lot behind Dec 04 18:10:51 poky $ git log --oneline 3cda6c0bd175282903079dea4affb3fcd290f137..origin/hardknott | wc -l Dec 04 18:10:55 987 Dec 04 18:11:18 and this change is one of them Dec 04 18:11:57 The baseline is latest revision in Yocto 3.3 Hardknott as of 2021-03-15. Dec 04 18:12:35 JaMa: Kirkstone I'm building here Dec 04 18:12:54 But seems it's trying to compile chromium-x11 and qtwebengine at the same time LOL Dec 04 18:13:06 That will quite likely give OOM Dec 04 18:16:38 "Tofe: hardknott has most of the results, but I would test kirkstone now" s/hardknott/zeus/g Dec 04 18:17:33 damn Ctrl-w in the wrong window Dec 04 18:17:44 looks like bb is now going on Dec 04 18:33:04 Tofe: just keep in mind that it might take whole day to finish Dec 04 18:33:28 Ok, so I didn't start it at the right time it seems Dec 04 18:34:05 you can let the fetch part to finish (as it doesn't block anything and can be easily restarted as many times as needed Dec 04 18:34:30 but before starting the 4-all, make sure you have something like 8 hours for just this one test Dec 04 18:34:45 Yes, there's only a couple of fetches to finish Dec 04 18:36:46 well less than 8, but stil maybe around 4 hours for each test 4,5,6,7 Dec 04 18:36:59 binutils will take the longest most likely Dec 04 18:37:11 15:12 < JaMa> it usually takes around 30 mins to fetch binutils from YP Dec 04 18:38:14 exactly Dec 04 19:04:48 "2763.00 seconds binutils-native-2.38-r0/do_fetch" damn Dec 04 19:09:26 damn, it's hard to find a good system monitor which is useful, light and traces the "right" things Dec 04 19:10:49 I don't want to install graphana just for monitoring cpu,ram,disk,network,and T° Dec 04 21:25:34 yeah, for run-time monitoring I'm using htop+atop, but for graphs and history I've ended with graphana and I'm still quite disappointed with it Dec 04 22:20:04 I found https://github.com/hakandundar34coding/system-monitoring-center today; a bit show-off, but it has all the data there. Still not really what I want, tough. Dec 04 22:20:07 +h **** ENDING LOGGING AT Mon Dec 05 02:59:56 2022