**** BEGIN LOGGING AT Wed May 08 02:59:56 2019 May 08 07:12:26 trying to build an SDK for core-image-minimal with latest master but it fails with: ERROR: core-image-minimal-1.0-r0 do_populate_sdk: [Errno 2] No such file or directory: '...tmp/work/qemux86-poky-linux/core-image-minimal/1.0-r0/sdk/image/opt/poky/2.7+snapshot/sysroots/x86_64-pokysdk-linux/usr/lib/locale' May 08 07:13:09 I've set USE_NLS = "no" and GLIBC_GENERATE_LOCALES="en_US" in local.conf. is this really not supported? May 08 08:37:09 tried to write a patch for the bug I filed https://bugzilla.yoctoproject.org/show_bug.cgi?id=13337 but native packages have so many differences to "normal" packages... May 08 08:37:10 Bug 13337: normal, Undecided, ---, richard.purdie, NEW , Error with rmt when image depends on both tar-native and cpio-native May 08 08:39:31 Neither sub-packages nor update-alternatives on native packages?! May 08 08:40:08 There's no native in update-alternatives. Oh wait, there actually is... :-D May 08 08:44:48 New news from stackoverflow: YOCTO - First build for BBB May 08 12:34:36 I'm seeing Python SSL: CERTIFICATE_VERIFY_FAILED while building a new recipe for gputop. I converted the recipe to inherit python3native, but the failure remains. Does this sound familiar? May 08 12:51:09 Does anyone know of a layer/recipe which provides syslinux as EFI_PROVIDER? Thanks! May 08 12:55:54 syslinux is in oe-core May 08 13:11:20 rburton: That recipes doesn't provide the EFI parts, unfortunately. May 08 13:11:36 alessioigor: patches welcome :) May 08 13:15:30 rburton: Right :-) May 08 13:38:23 rburton, latest webkit won't compile for x86 (32 bit), they hardcoded a check for sse2 May 08 13:38:51 is COMPATIBLE_HOST_i586 = 'null' correct then? May 08 13:38:57 (and same line for i686) May 08 14:19:22 ooo. fancy pr-tracker-bot on lkml May 08 14:33:47 RP: I was wondering how feasible it would be to hack an AUTOREV-for-HTTP entirely in metadata w/o hacking bitbake? I need a recipe to always fetch the latest file for CI purposes... Any pointers? Thanks! May 08 14:44:06 denix: I'd probably look up the parameter which renames the download and then do something like ;name=${@python call to randon filename generator} May 08 14:44:39 denix: would work on the basis that the name always changes. or just use DATETIME in there May 08 14:45:16 (this is in SRC_URI in case that isn;t clear) May 08 14:56:09 RP: but wouldn't always changing checksums pose a problem? May 08 14:57:06 denix: you'd need to disable the checksum checking May 08 15:03:06 RP: ok, got it, thanks! May 08 15:13:43 JPEW: around? May 08 15:13:52 RP: Yep May 08 15:14:10 JPEW: just had a play with the log dumping for resulttool. We may need a small tweak May 08 15:14:31 Sure May 08 15:14:32 JPEW: my test case happened to have the same ptest run several times, so it just overwrote the log file multiple times :/ May 08 15:15:00 JPEW: may need to check for existing file and append a number or something like that? May 08 15:15:14 not quite sure what the best approach is May 08 15:15:31 Ah, so I wasn't sure.... is that in multiple results objects? May 08 15:15:55 JPEW: yes, several test runs in one json file May 08 15:16:12 Can you send me the link to the file? May 08 15:17:17 JPEW: http://dan.rpsys.net/wp/rp/testresults.json.gz May 08 15:17:46 mmm, thats a big one May 08 15:18:49 JPEW: local debugging... May 08 15:20:50 Ya, ok. This is because multiple top level... results(? Needs a better name!) ran the same ptest May 08 15:21:15 e.g.: "runtime_core-image-sato_qemux86-64_20190508124620" May 08 15:22:36 JPEW: yes, I ran that test multiple times and it gets appended to the json each time May 08 15:22:54 JPEW: we'd not normally see this on the autobuilder but we do see it in local builds May 08 15:24:25 Got it. All my examples of the autobuilder output only had the one top level test, which would explain that.... I think there might be a few options, but I don't know which is best based on expected usage: May 08 15:25:02 JPEW: I hadn't expected this problem but my local build proved "interesting" :/ May 08 15:25:35 1) Append ptest results as you suggested, 2) make directories for each top level test run in the ptest output directory 3) Add a filter that allows the selection of a test run 4) always choose the "last" test run May 08 15:26:50 JPEW: I'd be tempted to do "if filename exists, try filename-1.log, then filename-2.log until it doesn't exist May 08 15:27:32 That doesn't associate test runs but I'm not sure that matters. Having the files separate would be useful for comparison May 08 15:27:46 Hmm, that could be annoying if you keep re-running the script. You'll get more and more files without really being able to figure out which are which May 08 15:28:12 true :/ May 08 15:28:32 We could add an option that does 2 if specified. If there is only one test run, it is optional, but is mandatory if there is more than one? May 08 15:29:10 JPEW: that would seem reasonable May 08 15:29:29 I'll give it a go and push up the patch to see if it looks ok May 08 15:30:02 JPEW: at the back of my mind is knowing side but side comparison is probably the most useful thing you could do with multiple files May 08 15:30:54 Ya, you could do a directory compare if you wanted... it's not as nice but would still work May 08 15:31:24 JPEW: yes, swings and roundabouts... May 08 15:32:11 RP: I think that coloquilism is lost on my uncultred midwestern ears :) May 08 15:37:29 JPEW: different and yet similar May 08 15:38:43 its defined online as "a situation in which different actions or options result in no eventual gain or loss.". I guess I thought it was in common use :) May 08 15:44:22 RP: I don't think I've ever heard it before :) May 08 16:17:23 RP: Patch sent.... I think I should probably also write some test case :) May 08 16:59:12 JPEW: thanks. I love testcases :) May 08 17:21:44 ugh, stuck waiting on two builds before i can get anything else done on these tasks, and i hate to juggle more than 2 tasks at a time or i'll start losing track of where i am in each.. May 08 17:22:02 thats the one thing i htae about doing oe development for a living, the test cycle times May 08 17:28:41 https://xkcd.com/303/ May 08 17:29:29 kergoth: I struggle with that a lot May 08 20:17:00 New news from stackoverflow: how do I know which package manager in installed on my linux [closed] May 08 20:38:20 I feel like there is some irony in ptest failing because it wasn't expecting to be successful at doing something May 08 20:39:07 Well, maybe not failing per se. May 08 20:39:23 JPEW: yes :) May 08 20:42:38 JPEW: that code makes me want to cry :( May 08 20:45:52 Eww.... gratituous getattr() is always a pain May 08 20:48:51 JPEW: its not entirely gratuitous, its all a bit convoluted though and there must be a better way May 08 20:49:19 well, I guess it is May 08 20:49:40 Thats fair. getattr() can be really useful, but it can easily go off the rails too :) May 08 21:19:04 JPEW: ptest log extraction is much nicer now thanks :) May 08 21:19:28 RP: Good May 08 21:21:56 Will that be sufficent to close https://bugzilla.yoctoproject.org/show_bug.cgi?id=13331 when it is on master? May 08 21:21:57 Bug 13331: normal, High, 2.8 M1, JPEWhacker, IN PROGRESS REVIEW , resulttool log command doesn't work as expected May 08 21:35:18 JPEW: yes, thanks! May 08 22:08:15 Hi all. I'm wondering if it's better to create a bbpappend to change SRC_URI to my fork or copy the recipe from the bsp. Or other other options I'm unaware about? May 08 22:16:52 prosper_: that depends whether its just that change or how many other changes you may need. How similar are the sources? May 08 22:21:25 i'm essentially creating my own board May 08 22:26:38 prosper_: probably into own recipe territory at that point but its just a guess May 08 22:38:36 ok thanks for your input May 08 22:40:02 actually the original bsp has a .inc that it include. Should I create a recipe that includes that? May 08 22:41:18 if it saves you duplicating things you don't need to... May 08 22:42:20 JPEW: well spotted with that thumb bug btw! ;) May 09 01:27:52 * armpit hmm, changes merged w/o review May 09 02:48:02 New news from stackoverflow: Yocto Rocko bitbake process reboot on Ubuntu **** ENDING LOGGING AT Thu May 09 02:59:56 2019