**** BEGIN LOGGING AT Tue Sep 08 02:59:58 2015 Sep 08 07:03:02 RP: testimage is giving quite grief to me http://paste.ubuntu.com/12312477/ is the log for run but you can see ssh to emulator is failing ( its ubuntu 14.04 host) When I launch core-image-sato manually ssh works ok, I am now trying to build it on another machine Sep 08 07:14:25 khem`: :( Sep 08 08:19:25 morning all Sep 08 08:42:12 good morning Sep 08 08:42:44 Hello everybody, I've a strange problem with yocto fido... Everytime I cancel the build or if I got an error during a bitbake, I'm unable to restart the build (I've to delete the cache in tmp folder) Sep 08 08:42:51 I got : ERROR: Command execution failed: Traceback (most recent call last): File "/home/swissdotnet/yocto/poky/bitbake/lib/bb/command.py", line 99, in runAsyncCommand self.cooker.updateCache() File "/home/swissdotnet/yocto/poky/bitbake/lib/bb/cooker.py", line 1468, in updateCache self.parser = CookerParser(self, filelist, masked) File "/home/swissdotnet/yocto/poky/bitbake/lib/bb/cooker.py", line 1907, in __init__ if n Sep 08 08:43:12 Piziwate: can you please use pastebin to show me the error? Sep 08 08:43:54 http://pastebin.com/eCfycGmb Sep 08 08:47:41 bluelightning: http://pastebin.com/eCfycGmb Sep 08 08:47:57 Piziwate: hmm, that's very strange... just looking into the code now Sep 08 08:48:58 bluelightning: Yes very strange ! I'm a little bit lost... Sep 08 08:49:15 it's a bug for sure, the question is what is triggering it Sep 08 08:49:50 bluelightning: I tried to run bitbacke with -DDD but this doesn't help... Sep 08 08:50:30 Piziwate: have you added your own recipes / layers to this configuration recently? Sep 08 08:50:53 also is this the latest on the fido branch? Sep 08 08:51:17 bluelightning: I use my layer, but it was working well with the daisy release... I updated the fido today based on the git repo Sep 08 09:07:14 bluelightning: If I remove my layer, this is working... but I don't understand what brakes the cache Sep 08 09:07:57 Piziwate: I'm not sure what could cause it either... there's nothing in your layer referring to the "file-checksums" varflag is there? Sep 08 09:09:22 bluelightning: ?? Sep 08 09:09:52 Piziwate: if you grep (or git grep) the entire layer for "file-checksums" are there any results? Sep 08 09:11:06 no Sep 08 09:12:57 somehow, invalid data is getting into the cache for that Sep 08 09:13:04 bluelightning: I think I got it !! Sep 08 09:13:10 oh really? Sep 08 09:14:14 bluelightning: I had a folder (test recipe) who was quickly renamed with spaces and special chars (like "fb-uc1701-mod (wip)"). I think he was not loving the space ! Sep 08 09:14:38 ok, at what level was that folder? Sep 08 09:15:03 bluelightning: In my meta repository under recipe-kernel Sep 08 09:15:13 ok, let me see if I can reproduce that Sep 08 09:18:54 well, I didn't get quite the same error but it still blew up Sep 08 09:18:58 I'll file a bug, thanks Sep 08 09:19:23 bluelightning: Thank you ! Sep 08 10:08:10 Bluelightening: Yocto uses SystemV by default - right? Sep 08 10:09:42 yes Sep 08 10:09:52 easy to flip to systemd if you want to use that instead Sep 08 10:10:20 Want to revert actually SystemD is too slow. Sep 08 10:13:08 raykinsella781: that sounds unlikely if everything is working as it should. What's your test case? Sep 08 10:13:55 Sorry - I am using a meta layer that explicitly substitutes SystemD for SystemV. Sep 08 10:15:22 raykinsella781: look in the docs for how to turn on systemd, look in the layer that does it, and either comment it out or reverse it in your local.conf Sep 08 10:30:59 Thank rburton Sep 08 15:50:03 kergoth: are you around? Sep 08 15:50:10 somewhat. whats up Sep 08 15:50:19 kergoth: that regex for the premirrors is causing a problem.. Sep 08 15:50:40 which regex? Sep 08 15:50:41 https://.*/.*/(.*) https://foo.com/\1 Sep 08 15:51:09 it's working, but then https://foo.com/\1 matches.. and adds the url again. Sep 08 15:51:24 it's actually more like: https://.*/.*/(.*) https://foo.com/a/\1 Sep 08 15:51:38 but I'm seeing https://foo.com/a/a/a/a/a/a/\1.. Sep 08 15:51:39 etc Sep 08 15:51:50 (log file eventually fills the disk as well) :) Sep 08 15:52:01 it seems like the premirrors are checking for matches in premirrors Sep 08 15:53:20 I suggested trying to play some regex games, using an intermediate value and then a final comparison.. but that may not be reasonable Sep 08 15:53:22 ah, yeah, we should be catching that recursion and halting it.. Sep 08 15:53:43 but as i suggested before, you should also be able to use https://.*/.*/ https://foo.com/ Sep 08 15:53:44 in theory Sep 08 15:53:53 it's worth noting the premirror that's adding has %3A instead of / Sep 08 15:53:56 not sure if that'll behave the same way or not offhand Sep 08 15:54:01 so it's no longer matching the end name Sep 08 15:55:18 so, the result of https://.*/.*/(.*) -> https://foo.com/a%3Abar -> https://foo.com/a%3Aa%3Abar, etc Sep 08 15:56:56 I'll try without the \1 Sep 08 15:57:19 matching the last component and then adding that component back seems a bit unnecessary, though i'm not sure it'll solve this problem regardless Sep 08 15:59:04 I will try Sep 08 15:59:18 although I was having isseus with cfengine-masterfile adding the version info in there somehow Sep 08 15:59:46 it did not solve the issue Sep 08 15:59:57 I suspect it's the %A3 Sep 08 16:00:49 oh wait, an issue with my testing.. Sep 08 16:08:01 nope, it's still causing the issue Sep 08 16:15:51 Okay, I have a solution Sep 08 16:16:04 and the problem isn't exactly what I described. Sep 08 16:16:38 The actual premirror is this: https://.*/.*/(.*) https://foo.com/${bar}:\1 Sep 08 16:16:47 So, bar keeps getting put in. Sep 08 16:16:59 if I do ${bar}:/\1, it'll work Sep 08 16:17:17 and I can (and have) fixed it on the server side by dropping the / off the : Sep 08 16:17:27 s/off/after/ Sep 08 16:19:58 I thought there was a slash, but the % encoding was throwing me off Sep 08 16:29:16 ah, interesting. i think we'll need to improve bitbake's recursive mirror application to do better at catching cases of recursion Sep 08 17:40:21 RP: so finally I managed to run the tests on a KDE system Sep 08 17:40:58 RP: and I could see the failure too. however when I eanbled kvm the failure went away Sep 08 17:41:37 some sort of timing Sep 08 17:41:47 I wonder if we are reading back the time too fast Sep 08 17:42:09 since manually I could never reproduce it that may speak of it. since manual inputs are delayed Sep 08 18:47:56 @JaMa , there's a specific channel to talk about meta-qt5? Sep 08 19:13:21 caiortp: #oe Sep 08 19:57:03 tks khem Sep 08 20:03:04 would anyone object if i rewrote the wget fetcher to use curl Sep 08 20:03:18 every time i need to use a proxy to get to the internet i curse wget for not supporting socks4a Sep 08 20:13:24 rburton: its a welcome change Sep 08 20:14:08 here is my prior attempt Sep 08 20:14:09 https://gist.github.com/kraj/4ef11aa4cc484c70f689 Sep 08 20:14:39 but its half done Sep 08 20:14:50 since its missing bitbake fetcher changes Sep 08 20:35:38 oh yay a good start though Sep 08 20:36:06 hang on —no-check-certificate, that's not good surely Sep 08 20:36:28 surely we should have that to check by default but allow some way to skip it if hosts are really stupid Sep 08 20:45:03 that does sound like it'd be a better bet Sep 08 21:06:29 caiortp: #oe channel is better (or openembedded-devel ML) Sep 08 21:06:46 tks ! Sep 08 21:10:13 weim Sep 09 02:23:24 has anyone make the time to look at x-compile in OE and http://mesonbuild.com ? It's been on my to do list but I won't get time for a long time... **** ENDING LOGGING AT Wed Sep 09 02:59:59 2015