18:01:20 #startmeeting 18:01:20 Meeting started Mon Mar 19 18:01:20 2012 UTC. The chair is bmwiedemann. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:01:20 Useful Commands: #action #agreed #help #info #idea #link #topic. 18:01:39 As always, today's agenda is on http://en.opensuse.org/openSUSE:Testing_meeting 18:01:39 You can still add to the agenda on the wiki while the meeting is in progress 18:01:49 # topic 12.2 testing 18:01:53 #topic 12.2 testing 18:03:00 MS2 was delayed for a week with openQA test-results showing poor quality until coolo fixed some bug 18:03:18 so now only the LiveCDs had problems installing 18:03:43 and for some reason, the consoletests ran in text mode instead of framebuffer 18:04:29 I did not do manual testing so far. 18:04:49 LWFinger: how was your experience so far? 18:05:48 I have installed both the NET Install (with an LXDE desktop) and the KDE Live CD, both on a VB VM. 18:06:48 The NET Install seemed to go OK, but I have not yet done much testing. I was mainly checking to see that the DHCP problem was still fixed. 18:08:06 The KDE Live CD had the GRUB2 problem with 2 missing files. I added the Bug # and the workaround to the most annoying bugs list. 18:09:01 great. 18:09:04 One other problem with KDE. No matter whether you select "Logg Off", "Restart", or "Shut Down", the effect is "Log Off". 18:09:22 yes, I also saw that one on openQA. 18:09:47 That is a minor annoyance as you still have to shut down or reboot from the Login screen. 18:10:02 Otherwise, no problems. 18:11:23 Done. 18:12:08 tigerfoot, mrdocs, Ilmehtar: do you have any testing experience to share? 18:12:27 One thing about the GRUB2 problem. Coolo was added to the Cc list for the bug report. It should be fixed soon. 18:12:58 bmwiedemann: not recently.. just following along for the moment.. been trying to fix gcc4.7 fails 18:13:38 mrdocs: also a good thing :) 18:13:38 Have you tried a kernel build with 4.7? 18:18:40 hmm... anything else to add to the 12.2 MS2 testing topic? 18:20:02 #topic Next Meeting 18:20:09 I just looked at the Beta forum to see what was reported there. One problem with "popping" in sound, but better than MS1. 18:20:40 One other user reported Video driver (Intel) problems. 18:20:53 could be pulseaudio or kernel probs 18:21:23 on factory ML there was also one report about Nvidia driver, which did not compile on MS2 (needed a patch) 18:22:31 That is always a problem - The patch I saw was not enough for my system with kernel 3.3. Nvidia is slow to react to new kernel API changes. 18:22:55 How about 2012-04-16 for the next meeting (would be 11 days after scheduled MS3 date)? 18:23:31 OK here. That would be income tax day in the US, but mine are already done. 18:24:16 Will Europe be on Summer Time by that date? 18:24:19 yes 18:24:35 so should we say 17:00 UTC then? 18:25:31 I think that is right. 18:26:08 good. I'll update it in the wiki after our meeting 18:26:13 #topic open discussion 18:26:20 sorry was in late ... 18:26:54 tigerfoot: Any MS2 experiences? 18:27:01 about factory, tittiatcoke made really nice progress with grub2/plymouth/systemd 18:27:46 we just have a pb with the boot luks.sh script but that should be solved in the next 10 days. (I need to unlock now my crypted lvm/btrfs) 18:27:48 was thinking: what are you guys thinking about how we do testing / QA for Factory. Is the overall quality sufficient? 18:28:25 bmwiedemann: absolutely not we miss lot's of case 18:28:41 it is definitely better with the automated things.. unfortunately you still need real people to find/fix things 18:28:50 I think so. I used the summary to tell when to start testing post-MS1 builds. 18:28:56 and I think real cases, but writing scenarios then convert them in automated tests is time consuming 18:29:11 * mrdocs noda 18:29:12 this time is not paid, then it didn't exist ... 18:29:15 nods even 18:29:48 but in the overall things to test and fixes, openQA is a wonderfull tools ... 18:30:21 we just need 10000 more tests, and find a way to run them quite easily on bare machine (ala Yast autoinstall) 18:30:21 Should oversize Live CD's be added to most-annoying bugs list? 18:30:26 the goal of openQA was not to ensure 100% bug-freeness, but to have something testable 18:30:49 any thoughts to auto-testing upgrades ? 18:30:50 eg 18:31:04 going from 12.1 > 12.2 ? 18:31:26 openQA is testing dist-upgrades both with zypper dup and from .iso 18:31:34 cool 18:31:34 just need to add 12.1 as a base-image 18:31:49 so for now it is only 11.3->12.2 and 11.4->12.2 18:32:23 Do those 11.X to 12.1 upgrades work? 18:32:36 this testing had caught a serious bug one day before 12.1 GM 18:32:59 wow 18:33:13 * tigerfoot remember ... 18:33:38 http://openqa.opensuse.org/viewimg/openqa/testresults/openSUSE-NET-i586-Build0263-11.4gnome32zdup/zdup-3.png looks like a bug from USRmerge 18:33:59 or something else? 18:34:36 but another zypper dup from 11.4kde worked 18:35:29 bmwiedemann: perhaps for the /usr merge, having a script that at least try to open each bash script and check shebang or dig inside them path about tools mis-using /bin /sbin 18:36:02 and another one finding all rpm who put binaries in /bin /sbin in place of symlink ? 18:36:08 IMHO, shell scripts should keep using /bin/sh 18:36:20 * tigerfoot doesn't speak perl that much ... 18:36:23 otherwise, you are loosing portability 18:36:34 How is that development coming? will /usr change be ready for 12.2? 18:36:42 bmwiedemann: shell script should use #!/usr/bin/env sh 18:36:55 really? no! 18:37:28 bmwiedemann: then your scripts normally are fully portable (even bsd etc .. with ksh, dash, bash etc ) 18:37:55 python or ruby should use also the env way, 18:37:55 you mean, those do not have /bin/sh ? 18:38:17 I found, that using env does make a slight difference 18:38:30 that is: you can not use killall NAMEOFYOURSCRIPT 18:39:26 because $0 of the process is set differently 18:41:13 bmwiedemann: oh really, was not aware of that ... 18:41:36 as for automated testing, I would like to start with those programs, that would impact most users if they failed 18:42:16 inkscape and gimp are already included ;) 18:42:42 maybe postfix & thunderbird could be worthwile 18:42:52 Firefox? 18:42:56 long done 18:43:21 firefox was about the first thing to test after xterm 18:44:21 (though, the firefox test is one of the hardest to maintain with the fast version updates / design changes and different theming for every of our 4 desktop environments) 18:47:18 The only other program I use very much is LibreOffice. 18:47:23 I think, I'll do auto-test of postfix+courier-imap+thunderbird next. covers a good part of the TCP/IP stack in the kernel, too 18:48:15 LWFinger: testing oowriter and oomath for now. are calc and impress important? 18:48:27 bmwiedemann: calc is 18:48:38 think about financial, etc ... 18:48:53 budget for openSUSE fundation :D 18:48:54 For me, calc is important. 18:49:08 OK, so I'll also do that soon 18:50:14 do we have any example tables with complicated macros? 18:52:57 A quick Google search did not show anything, but I found a thread that references their automated testing. Perhaps I can find something there. 18:53:04 I'll surely find or make me one 18:53:39 how about instant messaging programs? aren't they important, too? 18:53:55 bmwiedemann: poke mmeeks.. he might know how they do testing 18:54:34 I never do IM. 18:55:02 I also only do IRC for these meetings. 18:55:09 mrdocs: on last year's FOSDEM, I spoke to Mechtilde of openOffice QA, who told be about their testsuite, that can run core-tests in hours or extensive tests in days :) 18:56:17 OK. testing IM would be hard anyway, as we don't seem to have any IRC or Jabber servers that could be setup in the VM 18:56:54 ah, we have in OBS 19:00:16 any other ideas on what to improve how? 19:03:13 then we should conclude today's meeting 19:03:34 thank you all for attending and making this more than a chat between me and Larry :) 19:03:59 :) 19:04:08 #endmeeting