15:00:34 <adamw> #startmeeting Fedora QA meeting
15:00:34 <zodbot> Meeting started Mon Apr  7 15:00:34 2014 UTC.  The chair is adamw. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:34 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic.
15:00:39 <adamw> #meetingname fedora-qa
15:00:39 <zodbot> The meeting name has been set to 'fedora-qa'
15:00:43 <adamw> #Topic Roll call
15:00:44 <lnovich> thanks randomuser
15:00:47 <adamw> ahoyhoy, folks
15:00:50 <adamw> who's around?
15:00:52 * roshi is here
15:00:58 * satellit listening
15:01:01 <adamw> #chair roshi
15:01:01 <zodbot> Current chairs: adamw roshi
15:01:19 * somvannda can i listen too?
15:01:28 <roshi> for sure!
15:01:32 * pschindl is here
15:01:40 <somvannda> thanks
15:01:46 <roshi> you can even talk if you'd like to join in :)
15:01:57 <danofsatx-work> I'm here
15:01:59 <somvannda> okay, i will ask if i have question
15:02:16 <roshi> :)
15:02:17 <danofsatx-work> just flabbergasted that emacs is a dep to git-all
15:02:39 * cmurf here
15:02:46 * pwhalen lurks
15:03:09 * mkrizek is here
15:03:13 <adamw> danofsatx-work: ?
15:03:23 * Viking-Ice joins in
15:04:17 <danofsatx-work> adamw: nuthin, just complaining as usual. I don't like emacs, never have, never will. the git-all meta package includes emacs.
15:04:43 <adamw> oh, there's probably some plugin integration or something.
15:04:51 <roshi> very likely
15:04:57 * tflink is here
15:04:59 <danofsatx-work> yeah, likely
15:05:09 <adamw> #topic Previous meeting follow-up
15:05:16 <adamw> Just one action item from last time...
15:05:49 <adamw> #info "adamw to add closing tracker bugs post-release to https://fedoraproject.org/wiki/BugZappers/HouseKeeping/Trackers" - I did that, https://fedoraproject.org/w/index.php?title=BugZappers/HouseKeeping/Trackers&diff=374250&oldid=374249
15:06:06 <adamw> any other business from the last meeting or two we should wind up, that I forgot about?
15:06:27 * roshi has nothing
15:07:01 <danofsatx-work> that was a long, long time ago (in a galaxy far, far away)
15:08:20 <adamw> heh
15:08:23 <adamw> #topic Rawhide validation testing
15:08:46 <adamw> so, just wanted some time to kick around the plan for doing some testing on rawhide in the months before we get to alpha
15:09:04 <adamw> i say 'validation' but we're not exactly validating anything specific, just using the test plans we have to see if wheels are falling off the nightly builds
15:09:11 <danofsatx-work> have they fixed the nouveau bug yet?
15:09:13 * roshi just installed rawhide over 20 with boot.iso
15:09:19 <adamw> what nouveau bug?
15:09:25 <roshi> the lagginess of it maybe?
15:09:33 <danofsatx-work> hang on, let me find it agai
15:09:42 * roshi is on nouveau
15:09:52 <adamw> if anyone hasn't seen the proposal, it's https://lists.fedoraproject.org/pipermail/test/2014-April/120792.html
15:09:59 <adamw> there's nothing specifically laggy about nouveau i know of.
15:10:06 * kparal is late
15:10:08 <Viking-Ice> dont we need to rewrite the entire validation process with WG/Products in mind?
15:10:14 <adamw> rawhide just went to kernel 3.15rc0, which means all the debugging options are turned on, which makes it very slow.
15:10:24 <roshi> I just finished this install - so I need to poke at it before I say it's a bug
15:10:29 <adamw> Viking-Ice: i don't think that actually blocks doing some testing on rawhide
15:10:46 <adamw> Viking-Ice: if i just stick up the installation and base matrices, we're only in the base system anyway
15:10:52 <danofsatx-work> .bug 1044304
15:10:56 <zodbot> danofsatx-work: Bug 1044304 Segmentation fault in xorg-x11-drv-nouveau with kernel 3.13.0-0.rc3.git5.1.fc21.x86_64 - https://bugzilla.redhat.com/show_bug.cgi?id=1044304
15:10:59 <Viking-Ice> adamw, right
15:11:12 <adamw> roshi: if you want release-like performance, use a kernel from rawhide-nodebug. if you want something tolerable, boot with slub_debug=-
15:11:35 <roshi> good to know
15:11:49 <satellit> is there a link to tests?
15:12:06 <kparal> so, how are we going to deal with rawhide changing every day, but the validation matrix refreshing once a month (for example)?
15:12:22 <kparal> should we add some rawhide-specific instructions to it, related to this?
15:12:32 <adamw> danofsatx-work: we need you to check. that was never a general bug. or at least, i never hit it.
15:12:48 <roshi> from what adamw was saying, this is just sanity checks - so maybe put kernel version in the info
15:12:51 <adamw> kparal: yeah, i had an idea or two about that but didn't write anything in yet
15:12:57 <adamw> kparal: simplest option would just be to date the results
15:13:00 <danofsatx-work> yeah, just saw that....I'll grab an iso and try again tonight.
15:13:12 <kparal> adamw: put it directly into the cell?
15:13:14 <adamw> "PASS adamwill 04-07"
15:13:16 <adamw> yeah
15:13:21 <kparal> sounds reasonable
15:13:38 <kparal> we might need to adjust the legend, then
15:13:49 <adamw> we'd have to adjust the text a bit, yeah.
15:13:54 <roshi> sgtm
15:14:04 <adamw> and flag it up in the announcement, though people will soon see the first few entries and get the hang, i guess
15:14:13 <kparal> yeah
15:14:16 <roshi> true
15:14:17 <kparal> good idea
15:14:41 <adamw> Viking-Ice: we do need to figure out how we want to revamp the process for .next for sure, but i think we can do Installation/Base testing on rawhide without waiting for that first
15:15:20 <brunowolff> Does anaconda change every day? I thought the main purpose of this is to get more install testing. There are already a number of people running rawhide. But I don't think many of them do regular install testing.
15:15:35 <adamw> not every day, but fairly often
15:15:40 <adamw> and other things can change that affect the installer
15:15:42 <kparal> installation and desktop testing, I believe, both are important
15:15:54 <satellit> do we list the anaconda version also in results?
15:15:58 <adamw> kparal: in a general sense, or as part of this plan?
15:16:13 <kparal> adamw: desktop matrix is not in the plan?
15:16:13 <adamw> satellit: that might get a bit cramped; the nightly date should be sufficient info
15:16:13 <brunowolff> I think we already have informal desktop testing going on for rawhide.
15:16:24 <satellit> ok
15:16:49 <kparal> I somewhat assumed it would include desktop matrix, but we can start with just installation matrix
15:16:51 <Viking-Ice> adamw, what we will keep is the Installation/Base what the WG's will provide will be what they will add on top of that ( or deviate from it which most likely cloud WG will )
15:16:58 <adamw> kparal: i was suggesting that we leave it out for now in deference to viking's point about .next
15:17:23 <adamw> since we haven't decided what the overall plan for testing above the base layer is likely to be
15:17:39 <kparal> but still, some desktop testing can't hurt anyone :)
15:17:48 <satellit> boot.iso or lives or both?
15:18:13 <roshi> both, I would guess
15:18:24 <Viking-Ice> kparal, Honestly I dont know why we should they seem to be about to unleash gnome 3.12 update on GA releases
15:18:32 <Viking-Ice> without testing it first ( with us )
15:18:32 <cmurf> I've had difficulty finding rawhide builds
15:18:53 <adamw> cmurf: the links in the proposed page should be what you need
15:18:55 <kparal> yeah, we need to provide proper links in the announcement
15:19:10 <kparal> ah, maybe already done :)
15:19:12 <adamw> kparal: i already fixed the header of the page, where you're supposed to find links ;)
15:19:38 <cmurf> https://fedoraproject.org/wiki/Releases/Rawhide
15:19:39 <adamw> but we could maybe put them in the email for this, since we don't really have the issue of people confusing tcs/rcs with the 'final' alpha/beta/ga
15:19:57 <cmurf> That contains some 2000+ words without clearly saying where to get builds.
15:20:09 <adamw> well, it has a lot of ground to cover
15:20:23 <satellit> I d/l directly from koji...
15:20:30 <cmurf> Then I go to the nightlies link and nothing there gets me what I want in anything approximating an intuitive way.
15:21:13 <satellit> http://koji.fedoraproject.org/koji/tasks?start=0&state=all&view=flat&method=createLiveCD&order=-id
15:21:50 <cmurf> Yeah right, look through that and tell me what I click on next. Because I can't figure it out.
15:22:11 <cmurf> Basically I stumbled in the dark for 15 minutes or so and said fuck this shit. It was actually a really irritating experience.
15:22:31 <satellit> it lists the .iso after clicking on a sucessful build
15:22:32 * handsome_pirate stumbles in
15:22:33 <cmurf> So if there are only 100 people running Rawhide, I'd say, "umm yeah no shit".
15:22:42 * handsome_pirate has the bronchitis and stomach bug
15:22:45 <roshi> I'm going to be writing up a walkthrough/tutorial on getting the boot.iso and doing an install
15:23:02 <cmurf> Right well getting the boot.iso is easier, but even less intuitive.
15:23:35 <roshi> I can do nightlies as well - then take a look at a wiki page and go from there
15:23:42 <cmurf> I back up three steps, and find the instructions for pointing the installer to the rawhide URL
15:23:43 <cmurf> https://dl.fedoraproject.org/pub/fedora/linux/development/rawhide/x86_64/os/
15:23:46 <satellit> http://dl.fedoraproject.org/pub/fedora/linux/development/rawhide/x86_64/os/images/
15:23:52 <satellit> has boot .iso
15:24:13 <cmurf> And from there I stumble some more and find boot.iso which I just so happen to know what that is because in previous testing found out that netinstl.iso is boot.iso
15:24:24 <cmurf> Otherwise I'd still be blundering around for how to install Rawhide.
15:24:41 <nirik> if this is about the rawhide wiki page... it's a wiki. Improvements welcome.
15:24:44 <adamw> cmurf: sure, we can write it up a bit better
15:25:07 <adamw> cmurf: the nightlies link goes to koji because there's no handy directory full of nightlies any more, they just come out of koji, it's automated
15:25:27 <adamw> someone would have to do some unnecessary hand-work to maintain a list of all the last week or so of nightlies, and inevitably it wouldn't get updated properly
15:25:44 <nirik> we could construct something from fedmsg.
15:25:47 <adamw> someone could write a script to scrape koji or something, i guess, but...no-one has yet
15:25:48 <adamw> yeah, or that.
15:25:57 <cmurf> I understand the obscurity, but when people need a secret decoder ring to do something it's a humongous barrier to getting them to participate.
15:26:10 <roshi> true
15:26:16 <nirik> orr....
15:26:28 <nirik> looks like threebean already added it to the dash...
15:26:31 <nirik> https://apps.fedoraproject.org/releng-dash/
15:27:12 <nirik> so we could point to that instead.
15:27:17 <nirik> (now that it exists)
15:27:25 <cmurf> OK so maybe https://fedoraproject.org/wiki/Releases/Rawhide should point to that in the first, I don't know, 100 words?
15:27:26 <roshi> sweet
15:28:03 <satellit> nice page
15:28:05 <nirik> sure. It's not really been in existance very long... and I last reworked the rawhide page a year ago.
15:28:13 <cmurf> No I'm wrong it's still confusing.
15:28:26 <cmurf> click on Details, get a koji page with nothing to download
15:28:29 <roshi> you have to scroll down the poage to get there
15:28:40 <adamw> it's two links.
15:28:50 <cmurf> Fedora-Live-Desktop-x86_64  click on the details link
15:28:51 <satellit> .iso and .ks both work for lives and remixes
15:28:53 <danofsatx-work> You have to click on the Descendants link.
15:28:59 <adamw> nirik: could it link straight to the output rather than the Details page
15:29:05 <danofsatx-work> fwiw, I agree - it is totally 100% unintuitive.
15:29:08 <cmurf> oh got
15:29:12 <nirik> dunno.
15:29:13 <cmurf> t=d
15:29:15 <cmurf> there we go
15:29:27 <nirik> possibly, might need recoding work, not sure.
15:29:30 <cmurf> Pulling a tooth and not bleeding out is more intuitive.
15:30:25 <danofsatx-work> the (details) link should point to the details page. The "Fedora-Live-$spin-$arch"  should be a link to the iso
15:30:26 <nirik> yes, we are all going to die, it's hopeless. Lets just close out the meeting and all go home.
15:31:02 <cmurf> nirik: No that's not what I'm saying at all.
15:31:21 <nirik> danofsatx-work: yeah, seems reasonable.
15:31:37 <satellit> I just keep a link up in firefox and refresh it fot koji list...
15:31:39 <cmurf> I'm saying, it's not fun, it's irritating, it's like sand in the wrong places, it's a bad experience. Now if that's what we want, fine, mission accomplished. If we really want or need people to use Rawhide more often or in larger numbers something needs to change.
15:32:12 <adamw> cmurf: i don't think anyone's disagreeing, you just seem to be coming over as very angry and i'm not sure that's your intention.
15:32:36 <nirik> cmurf: I am saying that I agree we could make it better. You are just saying over and over it's horrible. It's horrible, we get it, can we try and make it better now?
15:32:55 <adamw> ^^^
15:33:09 <nirik> danofsatx-work: I'll suggest that change. Any others we could/should want?
15:33:37 <danofsatx-work> I'm looking through it now, just now grabbing the source to see if I notice anything
15:34:04 <cmurf> Well first answer the question if it really needs a better experience. If there's more validation testing that's going to happen, then probably yes we need to have a better experience.
15:34:13 <kparal> if we're going to link the releng dashboard, maybe include spinners on the initial page, so that people know something's happening? I saw an "empty" page for 10-15 seconds, then it filled up
15:34:15 <nirik> yes, I think so.
15:34:29 <danofsatx-work> yeah, I thought that too....and forgot
15:34:37 <nirik> kparal: yeah, it does a datagrepper query for the info
15:35:52 <danofsatx-work> oh wow, code is more in depth than I anticipated. will take me a bit - I'll ping you and/or threebean with more as I look at it.
15:36:24 <danofsatx-work> and hey, whaddya know, the releng meeting is going on next door as we speak
15:36:47 <nirik> https://github.com/fedora-infra/fedora-releng-dash/issues/5 and https://github.com/fedora-infra/fedora-releng-dash/issues/4 filed
15:36:53 <nirik> feel free to file others you want/need.
15:36:55 <kparal> thanks
15:37:49 <adamw> ahem
15:37:49 <adamw> so
15:37:56 <adamw> i think we've detoured far enough
15:38:02 <danofsatx-work> SQUIRREL!
15:38:08 <adamw> ooooooh squirrel?
15:38:58 <adamw> it sounds like we're mostly on board with the rawhide testing idea, so i'll refine all the bits that need refining (thanks for the notes everyone) and post up some initial matrices with an announcement email this week
15:38:58 <kparal> https://github.com/fedora-infra/fedora-releng-dash/issues/6 filed
15:39:24 <adamw> #action adamw to put out the initial rawhide validation testing matrices and announcement mails
15:39:35 <adamw> anything else on this topic?
15:40:05 <Viking-Ice> have the bits stabilized enough for validation testing
15:40:13 <satellit> all lives DE's to test ?
15:41:13 <adamw> Viking-Ice: that's sort of what we'll be finding out ;)
15:41:24 <adamw> satellit: for now i'll just put up the installation and base matrices
15:41:26 <Viking-Ice> adI'm a bit concern we waste peoples time in such early testing so at least major feature needs to have landed
15:41:29 <Viking-Ice> mean adamw
15:41:30 <satellit> k
15:41:39 <adamw> Viking-Ice: i've run a few installs with nightlies over the weeks, they've often worked
15:41:58 <adamw> i don't think anaconda team have major changes to land this cycle, they're currently on stabilization i believe
15:42:26 <Viking-Ice> adamw, not concern if they work or not what I'm saying we might be wasting peoples time of major distruptions have not landed yet in codes
15:42:54 <Viking-Ice> so we need to know that teams have moved to stabilizsation for the testing to be useful
15:42:58 <adamw> i get what you're saying, i think there's still value to the testing.
15:43:00 <Viking-Ice> ( like Anaconda apparenty has )
15:43:00 <danofsatx-work> meaning, why test Gnome 3.11 if 3.12 is on the way
15:43:09 <danofsatx-work> (for example)
15:43:26 <satellit> boot.iso has 3.12 now (last week)
15:43:40 <kparal> it's still valuable to find any issues with Rawhide desktop, because we need Rawhide to work well, in order to attract even more users to it
15:43:40 <Viking-Ice> right if we know that major,minor releases are underway
15:43:55 <Viking-Ice> kparal, why would we want to attract users to rawhide
15:44:00 <adamw> danofsatx-work: that would be a bad example, GNOME uses alternate release versioning: if you want to validate 3.12, 3.11 is *exactly* what you should test ;). but yes, I know what you're saying.
15:44:23 * danofsatx-work slinks away in shame
15:44:24 <kparal> Viking-Ice: because that would break the vicious circle of rawhide being broken because nobody uses it, therefore rawhide being broken
15:44:24 <Viking-Ice> kparal, you know as well as I do that only fraction of developers and maintainers pay attention to it
15:44:39 <kparal> Viking-Ice: now you nailed the reason
15:45:10 * nirik thinks more rawhide users is a good thing.
15:45:10 <Viking-Ice> kparal, testers are the wrong crowds + we know historically it's not useable until after alpha
15:45:47 <kparal> and the only way how to fix that is to have more rawhide users
15:45:58 <Viking-Ice> rawhide is the place for major disruptive changes
15:46:25 <Viking-Ice> if you get more people to use it you will eventually find your self having to invent another place for those major disruptive changes to land
15:46:44 <Viking-Ice> rawhide rawhide so to speak
15:47:04 <adamw> nope, that's not the idea at all.
15:47:12 <kparal> hmm, that's questionable. disruptive changes as long as they don't break the distro too much, or it's very well documented how to fix it (/usr move)
15:47:41 <Viking-Ice> adamw,  if the intent is to stabilizy rawhide by getting more users ( not maintainers ) to use it yes then it is the idea
15:47:59 <adamw> 'stabilize', like 'stable', is an overloaded term.
15:47:59 <kparal> I want Rawhide for bleeding edge power users, not just to be a package repo
15:48:05 <quickbooks> Some of the live cds in F20 cycle came with prelink on by default (except for Gnome); Aren't all the DE supposed to fall in line: https://fedorahosted.org/fesco/ticket/1183 ?
15:48:08 <adamw> no-one is proposing making rawhide more conservative.
15:48:20 <adamw> quickbooks: is this a QA issue?
15:48:24 <Viking-Ice> adamw, yes see kparal responce
15:48:28 <nirik> maintainers should be landing _usable_ changes in rawhide.
15:48:52 <Viking-Ice> maintainers should be doing a lot of things but in reality they arent
15:48:56 <adamw> anyway, this is veering miles off track again
15:48:59 <kparal> Viking-Ice: the state you were describing is the state we're trying to move away from
15:49:03 <roshi> Viking-Ice: the idea is to find bugs faster by having more people poking at rawhide
15:49:04 <adamw> and there's ten minutes left in themeeting
15:49:13 * kparal shuts up
15:49:14 <quickbooks> adamw, ok sorry
15:49:35 <adamw> i'll include a note in the announcement about this question. no-one's being required to do this testing; if you don't think it's useful, then certainly don't do it...
15:49:39 <adamw> quickbooks: that was a question :)
15:50:46 <adamw> is there something qa needs to worry about in that issue? or is it just for fesco?
15:50:57 <Viking-Ice> adamw, it's more us not asking until maintainers give green light that they are paying attention and now would be the time to start testing for the affected components in the validation matrix
15:51:17 <adamw> Viking-Ice: i already wrote that i ran it by anaconda team and they think it's a good idea. they're the main ones.
15:51:35 <adamw> kernel team always likes having reports from rawhide, we know that already.
15:51:59 <kparal> and the sooner you report it in gnome, the sooner it gets fixed
15:52:12 <Viking-Ice> adamw, then reporters making guess what's useful and what's not and participate based upon that
15:52:22 <Viking-Ice> so anaconda is safe you should perhaps note that as well as the kernel
15:52:22 <satellit> is it OK to file bugzilla for found problems? or will it clog up the system?
15:52:54 <adamw> satellit: the idea is to find major issues and file them in bz
15:52:57 <Viking-Ice> kparal, we dont have 3.14 which I think will end up being released for F21
15:53:00 <satellit> k
15:53:22 <kparal> this is rawhide, not f21 testing. for the moment
15:53:25 <Viking-Ice> rawhide is on 3.12 ( but if that is final sure more testing the better )
15:54:00 <adamw> #topic open floor
15:54:10 <adamw> before we run over time, was anyone holding anything else for open floor?
15:54:31 <adamw> danofsatx-work: before I forget - on your nouveau issue you can just install a rawhide kernel on f20 and see if it works, that's usually not a problem.
15:55:00 <danofsatx-work> yeah, but I can't do it until this evening at the earliest.
15:55:36 <Viking-Ice> well the 3.12 being pushed to final is a bit alarming
15:55:42 <Viking-Ice> mean not final but GA releases
15:56:11 <Viking-Ice> based on some vague testing from hugsies repo
15:56:48 <nirik> nothing has been pushed yet. They asked for testing feedback before bringing the exception to fesco.
15:56:59 <adamw> i don't think it's necessarily a bad idea, but it has to go through fesco anyway.
15:57:05 <roshi> I've got some more on test maps - but was planning to respond to the list
15:57:21 <nirik> and the testing in fact did uncover some more things. :)
15:57:35 <adamw> #info Desktop team is considering submitted an exception request to have GNOME 3.12 as an update for Fedora 20, see https://lists.fedoraproject.org/pipermail/desktop/2014-April/009575.html
15:58:07 <adamw> yeah, it's not exactly been 'some vague testing', they made a copr expressly for people to run/test 3.12 on F20 and publicized it pretty widely.
15:58:19 <Viking-Ice> well the gnomers are known for they ui stability and a simply removal of the network icon in the top panel caused uproar for admins on these parts since easy access to their vpn profiles vanished when docket etc
15:58:30 <nirik> the upower thing needs dealing with for sure before they go any further.
15:59:20 <adamw> 3.12 already put that back...
15:59:20 <Viking-Ice> in anycase we need to map out the changes in ui no matter how small the devs think they are users might not agree suddenly finding them on 3.12 and maybe not being able to downgrade from it
15:59:49 <adamw> sure
16:01:19 <adamw> OK, that's our hour...
16:01:23 * adamw sets the quantum fuse
16:02:03 <roshi> ...another universe destroyed
16:03:03 <adamw> whoopsie
16:03:52 <adamw> thanks for coming, folks!
16:03:57 <adamw> #endmeeting