--- Log opened Thu Jan 25 11:03:22 2007
Lovechildand one man can only watch so much porn (and HD movie trailers)11:03
@wwoodsHAH. that's the first line of the meeting log, now11:03
Lovechildanyone from the QA team going to FOSDEM?11:04
-!- dmalcolm changed the topic of #fedora-qa to: Fedora QA | Meeting starting now! | <Lovechild> and one man can only watch so much porn (and HD movie trailers)11:04
@wwoodshee hee!11:04
LovechildI hate you 11:04
-!- dmalcolm changed the topic of #fedora-qa to: Fedora QA | Meeting starting now!11:04
@wwoodsLovechild: but the fluendo codecs all basically work? that's really cool11:04
Lovechildwwoods: using DECODEBIN2 and GStreamer CVS I can play pretty much every WMV/WMV file including the HD content11:05
@wwoodsLovechild: ooh, FOSDEM would have been cool, but it's a month away? I'm not sure I could get permission/budget/cheap tickets11:05
-!- leenuxg33k [i=leenuxg3@nat/redhat/x-d61127e197dc1c4e] has joined #fedora-qa11:05
LovechildA friend of mine offered to pay the busticket down there so I figured I might go and flash the Fedora love11:05
@wwoodsLovechild: oh nice!11:06
@wwoodspoelcat: ping11:06
poelcatwwoods: present and accounted for11:06
Lovechildalso Development now boots on dmraid and Compiz got fixed.. outside of Evolution which is always broken Development is looking solid from here11:06
@wwoodsLovechild: which compiz issue? the only problem I've seen is the window-placement badness11:07
@wwoodswell, actually, hold that thought a sec - let's get started with this meeting11:07
-!- wwoods changed the topic of #fedora-qa to: Fedora QA | Meeting starting now! | Agenda: http://fedoraproject.org/wiki/QA/Meetings/2007012511:07
@wwoodsHello and welcome and all that! The agenda is on the wiki as always (http://fedoraproject.org/wiki/QA/Meetings/20070125 if you missed the topic)11:08
[Users #fedora-qa]11:08
[@wwoods ] [ c4chris__] [ jhutar ] [ mether ] [ Oldma1 ] [ Sonar_Guy] 11:08
[ BobJensen-Away] [ dmalcolm ] [ leenuxg33k] [ mspevack ] [ poelcat] [ thl ] 11:08
[ bpepple ] [ Foolish ] [ lmacken ] [ mutk ] [ pyxel ] 11:08
[ bress ] [ giallu ] [ Lovechild ] [ nman64_away] [ sangu ] 11:08
-!- Irssi: #fedora-qa: Total of 22 nicks [1 ops, 0 halfops, 0 voices, 21 normal]11:08
Lovechildall hail the mighty Will Woods and all that11:08
@wwoodswho's actually here?11:08
LovechildDavid Nielsen present and accounted for11:09
Oldma1Roll Call?11:09
* lmacken is kind of sort of not really here11:09
@wwoodsLovechild: do I have your email address? I think I had asked you for a bit about the Love Day stuff (And a convenient excuse to get your email address for future use)11:09
Lovechildwwoods: david@lovesunix.net11:10
@wwoodsLovechild: gotcha, thanks11:10
Lovechildyeah I' m behind schedule on that massively11:10
@wwoodsOldma1: sure, mostly I just want to see how many people are actively listening and how many are idle 11:10
@wwoodsLovechild: s'ok, we got Test1 to worry about right now11:10
@wwoodsbut Test1 is a good place to start hashing this stuff out11:11
@wwoodsLovechild's obviously testing it, who else has attempted an install or otherwise messed with it?11:11
@wwoodsit's frozen now (yaaaay) so what's currently in development will be released on Tuesday as Test111:12
@wwoodsso we must do our best to make sure there aren't Big Nasty Bugs in it11:12
@wwoodsthat's the test matrix11:12
@wwoodsthat's all the stuff we should test in the next week11:12
@wwoodsif anyone doesn't have wiki edit rights, you can talk to me or mether or poelstra to get updates made11:13
@wwoodsoh, I should fix the "Current Testing Tree" thing11:13
@wwoodsand probably we'll need to add some stuff to the matrix about the LiveCD11:13
@wwoodsby the way - the LiveCD is installable now! it's pretty dang cool.11:14
Lovechildis that davidz' doing?11:14
@wwoodsyeah davidz + jeremy (katz) 11:14
Lovechildthey are nay as much men as gods11:14
Lovechildso who gets to fix Pup?11:15
@wwoodsas such we must make offerings of time and sweat unto them11:15
@wwoodsLovechild: oh, the comps badness?11:15
Lovechildwwoods: no but it still locks the GUI so it looks like it crashed - you would not believe the amount of shit I get for that one11:15
@wwoodsLovechild: hmm, that's just a general pup problem though, isn't it? 11:15
metherhave anyone started submitting the testing tools to the package review process?11:16
@wwoodsthe fact that it's not multithreaded or something, so the screen doesn't get redrawn sometimes11:16
@wwoodsmether: yep! dmalcolm submitted them yesterday11:16
metherwwoods: yes. its generally that Pirut, Pup etc dont provide enough feedback during esp during dep resolving 11:16
Lovechildit's a "developer hates threads and wants to make users suffer because of it" issue11:16
@wwoodsmether: we'll talk about that right after test111:16
@wwoodsLovechild: ha11:16
metherwwoods: how is test1 looking like now11:16
dmalcolmmether: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=22427111:17
@wwoodsfrom what I've seen it runs pretty well, although I haven't attempted an install since just before the freeze11:17
dmalcolmmether: Bugzilla Bug 224271: Review Request: rhts - A system for developing automated tests11:17
metherdmalcolm: do you really want to call it rhts?11:17
metherI thought we had a name like beaker or something11:17
@wwoodsLovechild: you can confirm it's installable on i386? how about x86_64/ppc?11:17
LovechildLovechild: x86_64 tested via upgrade from FC6 (slight issues with repo which must not be named)11:18
LovechildI'll do i386 in the morrow11:18
@wwoodsmether: it's kind of complicated, but as I see it, RHTS is the (fairly basic) stuff that Red Hat is providing - the stuff we build on top of it will be called Beaker11:19
Lovechildas well as clean net installs, the dmraid raid now boots which makes testing much easier11:19
metherwwoods: when QA meetings are over, please do send a summary and link to logs to folks in fedora-test list11:19
@wwoodsit *could* end up being the case that we replace the rhts stuff inside beaker, and drop the rhts name altogether11:19
metherwwoods: ok. 11:19
@wwoodsmether: yes, I forgot to do that last week. Mea culpa, won't happen this time11:19
@wwoodsLovechild: very cool. Do you have wiki edit rights? 11:20
Lovechildwwoods: I don't think so my user is DavidNielsen (how original)11:20
metherhow many tests are in beaker now? when do we expect to start running those tests on a regular basis? have we decided which list to send them to?11:20
@wwoodsLovechild: oh, can you do a 'yum list extras' after the upgrade and see if there's anything funky there?11:21
@wwoodsmether: I'm not sure about the unstable ('sandbox') tests, but there's 2 promoted tests in the upstream repo.11:21
@wwoodsmether: I'm hoping we can kickstart the test repo during FUDCon11:22
Lovechildthe upgrade is not all that smooth but it's perfectly workable11:22
metherdo we plan to start "supporting" test to final releases or F7 to F8 upgrades using yum?11:22
Lovechildthat would be madness to officially support that, unofficially it might work.. no?11:23
metheror a upgrade process decoupled from Anaconda but not necessarily directly with yum11:23
@wwoodsmether: that's a good question. do you think it's reasonable to bring that to the devs as a requirement?11:23
@wwoodsI think test3->final is reasonable to support11:24
Lovechildthe test releases aren't stable, nobody installs a test release without applying updates from Development - in effect we'd be supporting upgrade from any point in Development to Final fully tested11:24
@wwoodsLovechild: right, which is a dicey proposition11:24
Lovechildtest3 might work11:24
metherIf the update tool, refuses to put out updates without a proper update path for individual packages and checks to make sure that all dependencies are resolved, we all agree that epochs are not all that evil if it gets the job done11:25
metherthen yes i do think11:25
@wwoodstest3 is after the feature freeze and a bunch of stabilization, so officially supporting test3->final is reasonable in my book11:25
metherit is very reasonable to support it in a official capacity in most scenarios11:25
Lovechildif people know that the upgrade from test3 to final is supported, there's a bigger chance we can get more people on there to hit more issues - instead of getting a shit storm when final hits and it doesn't boot on half the machines in the world11:26
metherbasically yes, once we put out a rpm into the development or stable tree, there should always be a guaranteed update path 11:26
@wwoodsI agree, but that proposition depends on some things that don't quite exist yet - magic dep checking for the updates tool, for instance11:26
@wwoodsLovechild: right11:26
@wwoodsand AFAIK there's never been any big problems with Test3->Final upgrades11:26
@wwoodsif we are careful about taking note of any potential issues with test3->final, and pass those along to the anaconda folks11:27
@wwoodsso if they need special handling, they can do it11:27
LovechildI've always done rolling upgrades all through Development, it has never been that much of a problem and if we test it heavily it's doable11:27
@wwoodsthen I think we can officially support test3->final without putting undue strain on anyone11:27
LovechildI think it's worthwhile, I'd at least give time towards such testing11:28
metheris that feasible from F7 test2 onwards. can we try?11:28
@wwoodsmether: try what - officially supporting upgrades from test2->final?11:28
metherdo we know how close is Luke is to finishing the new update tools with evr and dep checks ?11:28
metherwwoods: yes11:28
Lovechildin this cycle could we limit it to test3, I think we are already taking on a lot of testing with tools are aren't yet ready and a small team11:29
metherwell test2-test3 upgrades should show up problems 11:29
metherwe can learn and fix11:29
metheryes, that generally is more work but F7 is ambitious release in general11:30
@wwoodsyeah, I think we should propose official support for test3->final upgrades, and *possibly* test1->test2 and test2->test311:30
Lovechilddid the Unity people offer up some preliminary test1 isos for me to throw at my setup?11:30
metherso we are risking failures in some points on not meeting goals which is ok IMO11:30
metherwwoods: if so the test release announcements should state that prominently  11:31
lmackenmether: I am going to port the old dep checking code over to the new tool this weekend11:31
@wwoodsLovechild: I'm asking f13 about iso sets11:31
lmackenwhich should be fairly trivial11:31
metherlmacken, rpm evr checks?11:31
lmackenmether: what kind of checks? broken update paths ?11:32
metherlmacken, yes precisely11:32
lmackenmether: i plan on doing that this weekend too. (it's probably a one-liner (rpm.labelCompare or something))11:32
metherlmacken, the update should just refuse to put out updates like that in all branches 11:32
metherincluding the development tree11:32
methergreat. when do we expect the new update system to be in action?11:33
lmackenmether: that would require the updates system access to the development tree?11:33
metherlmacken, yes. is that a problem?11:33
lmackenhopefully soon!11:33
-!- f13 [i=jkeating@fedora/ender] has joined #fedora-qa11:33
lmackenmether: not really a problem.  I'm sure we'll figure something out11:34
@wwoodsf13: howdy11:34
f13hi all11:34
Lovechildall hail the mighty Jesse Keating11:34
metherf13: Hello. how many images would the default installation of desktop spin for test1 be11:34
f13i386:3 x86_64/ppc: 411:34
f13unless something durastically changes.11:35
metherf13: Would it be possible to cut it down to 2 for the default on all arches (just for the defaults) ?11:35
f13oh wait, the default install11:36
Lovechild4 cds for a desktop system seems excessive11:36
f13sorry, I haven't actually tried the default install with CD isos11:36
f13the whole spin is 3 CDs11:36
f13Lovechild: tell the world to get rid of more languages.11:36
f13Lovechild: seriously, oo.org+languages is 1.5CDs alone11:36
Lovechildf13: I'll start with my own11:36
LovechildI hate OOo11:37
metherf13: Would it be reasonable to 2 CD's as a goal for the defaults?11:37
f13mether: once I get a spin that actually has the right content, I'll burn isos and see how many it takes to install.11:37
f13mether: personally, I don't really want to play that game right now.11:37
metherf13: ok. Its not really a game as much as getting more people to try out things easily 11:37
@wwoodsmether: well, the liveCD should help with some of that11:38
metherwwoods: true11:38
f13mether: thats why there is a LiveCD11:38
f13you can't just say "You only need the first 2 CDs" as that only works for English11:38
f13if you're !English, you're going ot need those lang packs, which will be on other CDs11:38
metherwell there is a reason why I explicitly mention defaults11:38
f13mether: default for whom?11:39
@wwoodsf13: oh hey, syncing out the liveCDs should also be way quicker - are you building those too, or is that all jeremy?11:39
metherdefault in comps.xml11:39
f13default changes depending on what language you are, what arch you're running, etc..11:39
f13wwoods: jeremy has graciously taken on composing the LiveCD set for this release.11:39
Lovechildf13: when can we expect test1 test iso ?11:39
metherEnglish as default then11:39
f13Lovechild: I hope to have some RC isos out later today somewhere, with unsigned packages.11:39
f13need to have the final test1 isos ready by sometime on Friday for a Tuesday release11:40
f13our IS needs a bit of time to stage, and mirrors need time to sync11:40
metherDavid Zeuthen was talking about integrating live cd's into the development process and even have live cd respins post release on a regular basis11:40
Lovechildf13: as a special favor to me, if hal detects a Plextor drive can you offer me a button to stab the Plextor CEO in the face?11:40
metherLovechild: i can see you are very passionate 11:41
Lovechildonce again burning died because of their proprietary extensions11:41
@wwoodsf13: so here's the funky bit - installer testing should happen *before* the release, but we need trees/iso sets to test with 11:41
f13mether: I haven't integrated the liveCD creator into pungi, so not this time around.11:41
metherbefore F7 GA?11:41
f13wwoods: hey guess what, people can compose at home!  (:11:41
f13mether: more than likely11:41
@wwoodsf13: ooooh hey, good point11:41
Lovechildmether: when I pay 3 times the price of a cheap dvd burner that claims to have perfect Linux support and I see nothing but pain for it.. I get passionate11:42
f13mether: right now things are an odd mix of distill, extras push tools, pungi, and livecd-creator11:42
@wwoodsf13: can you give me (or us) some really quick instructions on how to compose media from rawhide?11:42
f13wwoods: I started a page on how to use pungi in mock, I'm going to post up the config files I'm using once I'm happy with the content we're getting (should know very shortly)11:42
* wwoods hoping the answer is: "I already wrote them up, they're here on the wiki, haven't you seen them already you dumbass?"11:42
-!- pyxel [n=pyxel@] has quit ["can be the farce with you, old man !"]11:42
metherf13: I was expecting to have a bit of a mess and churn due to the core+extras merge11:42
@wwoodsf13: ossum11:42
f13wwoods: I've been doing some hopefully good documentation of pungi as I go.11:42
metherf13: whats the status on brew?11:43
f13mether: once we merge, and start looking at how we create rawhide nad such differently, then we can start doing automated liveCD type stuff.11:43
f13mether: should have legal resolved today, as you heard in the board meeting yesterday11:43
metherf13: that would very very cool11:43
metherf13: I missed out some details in the last board meetings. I was a bit late11:43
metherf13: and there was other important things of higher priority to discuss11:44
f13mether: sure.11:44
f13wwoods: https://hosted.fedoraproject.org/projects/pungi/wiki/PungiDocs is the start of docs11:45
-!- Bob-Laptop [n=Robert@fedora/pdpc.sustaining.BobJensen] has joined #Fedora-QA11:45
metherf13: why does number of CD's change so much between arch's?11:45
f13https://hosted.fedoraproject.org/projects/pungi/wiki/PungiDocs/RunningPungi and https://hosted.fedoraproject.org/projects/pungi/wiki/PungiDocs/RunningPungiInMock is where more meat will show up11:45
f13mether: because x86_64/ppc composes are multilib.11:45
@wwoodsf13: beautiful. So I guess our plan for testing the test-releases will be: after freeze, it's up to us testers to build iso sets and install from them11:45
f13mether: x86_64 includes a variety of i386 packages, ppc includes ppc64 packages.11:45
@wwoodssince (in theory) we're building by the same process as the official one, the test results should be fine either way11:46
f13wwoods: right11:46
f13wwoods: I'll be spinning locally, and probably at the same time, spinning in phoenix where the world can get to it.11:46
metherf13: is what Jonathan Corbet pointed out fedora-test list a RPM  multilib bug? It seems strange to have a default behavior like that 11:46
f13mether: you're going to have to be more specific.11:47
Lovechildmultilib makes kittens cry11:47
metherf13: https://www.redhat.com/archives/fedora-test-list/2007-January/msg00263.html11:48
LovechildI think that would be the multiple packages owning the same files issue right?11:48
f13yes, there are some multilib rpm bugs.  Yes we'd like to fix them.  No we haven't yet hired anybody to help Paul Nasrat with rpm AFAIK, and unfortunately Paul has been very busy with RHEL5 issues.11:48
@wwoodswas that the 'fighting the i386 plague' thread?11:48
metherwwoods: yes11:48
@wwoodsinstalling both arches by default is the intended behavior, as I understand it11:48
@wwoodsit seems odd but it makes multilib Just Work, which is what people want11:49
f13mether: I still need to actually verify that scenario, it could actually be a yum bug.11:49
@wwoodsso following the principal of least surprise..11:49
metherwwoods: well, saying we install both arches by default doesnt actually answer that question if you read the discussion 11:49
f13install both by default yes, however if replacing an obsoleted package, and only obsoleted.x86_64 was installed, new.i386 shouldn't also be installed IIRC11:49
@wwoodsmultilib makes my soul hurt11:50
f13obso.x86_64 -> new.i386 + new.x86_64 sounds like a bug, but I want to talk with Jeremy and Seth more about it.11:50
f13might just need a fine tune of the depsolving logic in yum11:50
f13and yes, multilib makes baby jesus cry11:50
@wwoodsf13: do you think it's reasonable to officially support test3->final upgrades?11:50
f13wwoods: *shrug*  I question reasonablness of ANY upgrades.11:51
@wwoods(probably with a "there may be some manual work involved but it *will* work" caveat)11:51
LovechildI have yet to have one person say "Multilib is great.. I love you guys for forcing it upon me"11:51
f13that said, I could see just a policy of ensuring nvr upgrade paths from test3 on11:51
@wwoodsso what you're saying is that it's no more risky than any other upgrade? heh11:51
f13other than that... um...  11:51
@wwoodsf13: right, lmacken was saying that the new updates tool was going to help enforce nvr sanity11:52
f13see, Anaconda has black magic in it for doing upgrades, to take you from one state to the next.  ALl that magic expects the state to be N-1, not Ntest311:52
f13so, while nvr sanity can be enforced, that doesn't help the black magic stuff.  It _should_ mostly work, but I'm not laying bets down.11:53
@wwoodsso maybe this is something I should discuss with jeremy et. al. as well?11:53
f13well, the answer your'e going to get is " there is only so much time in a release.  We can concentrate on making foo -> bar work, but most likely not foo+test3 -> bar, or just test3 -> bar"11:53
methermaybe we shouldnt have such black magic and make such assumptions?11:53
f13mether: then we'll never be able to make significant changes between releases.11:53
f13things that need an non-root filesystem11:54
metherf13: It is possible to have a separate upgrade tool do some changes if necessary11:54
f13we turn into Debian, sitting on kernel 2.6 until after 2.8 is released11:54
@wwoodsthere are unavoidable weirdnesses that must happen for major upgrades, but doing test3->final should be a minor upgrade .. it's after feature-freeze, if memory serves11:54
f13wwoods: yes.11:54
metherf13: Umm. That has nothing to do with enabling a proper upgrade11:54
f13wwoods: I think the best we can do is enforce nvr paths, after that its.... *shrug*11:55
@wwoodsbut anaconda is designed to do major upgrades11:55
@wwoodsso it doesn't necessarily know what the Right Thing To Do is for a minor one.. hmm11:55
f13mether: anaconda "black magic" is necessary to take a 2.4 userland and translate it into a 2.6 userland, staticdev to udev, lvm to lvm2, etc...11:55
Lovechildsounds like platform upgrades is best handled by a specific little tool, I see a lot of requests to support yum updates from say FC6 to F7 without the need to download a DVD image11:55
metherwwoods: 2.4 to 2.6 live upgrade does work on debian using apt-get11:56
f13mether: they're not doing some of the other changes.11:56
Lovechildwait.. debian has releases?11:56
metherf13: we cant do a lvm1 to lvm2 conversion on the fly and thats beyond the scope of yum but those kind of changes are few and far between11:56
metherthose changes can be encapsulated in a separate upgrade tool hooked into pup and puplet 11:57
f13mether: yes, but those are the types of changes that require blackmagic, and require us to have the freedom to MAKE such changes between releases.11:57
f13those changes require you to reboot and do operations with your root file system offline.11:57
Lovechildmether: one wonders how much work such a tool would be to write vs. the benefits11:57
metherDo folks realize that Ubuntu is trying 11:58
f13because that tool _is_ anaconda11:58
@wwoodsit seems like you could probably chain together yumdownloader with the pxeboot vmlinuz/initrd11:58
@wwoodsso you fetch all the packages to update, grab the boot images11:58
f13wwoods: use boot.iso, point to the remote repo, you'll only download the newer packages, and still get the black magic.11:58
methercan I call anaconda from pup/puplet? 11:58
Lovechildmether: yes and they are also moving towards making them able to convert a Windows desktop to a Linux desktop, files intact and all.. very cool11:58
@wwoodsf13: right, it would be cool if the downloading happened before you booted into anaconda11:58
f13wwoods: local mirror... la la la11:59
metherLovechild: there is no reason for us to be not even trying11:59
@wwoodsf13: hah11:59
metherLovechild: its better than the pain of supporting a release for many many years11:59
metherit encourages more folks to upgrade 11:59
f13uh...  Fedora doesn't do that11:59
Lovechildmether: I never said it wasn't.. I just implied that we might want to poke them for cooperation - a lot of the pieces would be common11:59
@wwoodswell, we'll need a lot more discussion about this stuff, but I don't know if it's gonna happen for F711:59
metherLovechild: the pieces are easily separated, yes people are looking into that already12:00
Lovechildmether: you just made my day12:00
@wwoodsobviously it bears talking to the anaconda folks about how hard it is to support test3->final12:00
metherLovechild: its also the reason why I harped on ntfs stuff being integrated into the release 12:00
@wwoodsbut I think we're off on a tangent here and I'd like to get back to the agenda (and RHTS)12:00
Lovechildmether: yes and now that legal allows it there shouldn't be a problem.. yay for legal12:00
metherLovechild: not quite but thats a different discussion12:01
f13wwoods: so, I think the nvr enforcement is the easily doable part12:01
@wwoodsf13: right, given that test3 is post-feature-freeze there shouldn't be any black magic12:01
f13of course, some will say we should do it for ALL of development nad never break NVRs, but....12:01
metherf13: did PC agree on using epochs if necessary?12:02
Lovechildf13: one step at the time12:02
f13mether: that went down a scary path of doom12:02
metheras in there is no hope?12:02
metherwe just need to have that discussion anyhow12:02
f13I had no hope of that particular thead12:02
@wwoodsI'd be happy with test3->final for F7. we can revisit the topic some other time.12:03
f13but perhaps a "use epoch only as a last resort" would probably pass12:03
metherf13: what is a last resort needs guidelines 12:03
metherits a key to ensuring smoother upgrades12:04
f13"When there is no other way to make the upgrade path work, force it with an epoch"12:04
f13pretty simple.12:04
metherwell that discussion in packaging list didnt give me that impression12:04
f13I'm FAR more interested in developming tools to help migrade user settings or homedirs and seeing more fresh installs.12:04
metherit seems there is a lot of contention about the usage of epochs and when it is advisable or even acceptable  to do so12:04
@wwoodsyeah, something that magically saved all your Stuff and did a fresh install would be rad12:05
f13upgrades are just a poor experience.  YOu don't get the new themes, you don't get the new settings, you don't get to pick new packages, etc...12:05
f13mether: welcome to the packaging commitee.12:05
Bob-LaptopI agree f1312:05
metherf13: well thats the whole point. I want to retain the same things while getting a refresh of my existing packages 12:05
f13packaging is so completely freeform it is extremely difficult to get folks to agree upon a best practice.12:05
metherIf i want the world to change on every upgrade I might as well as do a fresh installation 12:06
Lovechildwhat that one one reason to merger core/extras?12:06
f13mether: so then why are you doing an upgrade at all?12:06
f13mether: we're still giving you updates to your rleease12:06
metherf13: because my current release is EOL'ed?12:06
f13so new stuff every 13 months is too soon for you?12:06
metherbecause I want new features only available in the next release12:06
Lovechildaren't we getting close to the top of the our?12:06
@wwoodsLovechild: yeah12:06
f13mether: but you might not GET those new features via an upgrade12:07
f13if you didn't have compiz before, you won't have compiz after the install.12:07
@wwoodsmether, f13: I think the upgrade discussion is beyond the scope of testing, at this point12:07
f13wwoods: agreed, sorry for taking over there a bit :?12:07
Lovechildf13: world domination all in due time my friend12:07
@wwoodss'ok, mostly I just want to talk about exciting RHTS stuff (and go get lunch)12:07
@wwoodsthe upgrade discussion is a really important one12:08
@wwoodsbut.. some other time12:08
metherfine by me. carry on. I thought I wanted to clear up our stand point before t1. 12:08
@wwoodsas far as I can see, Upgrades are going to stay as they are for F7, but we are going to make a concerted effort to keep nvr sanity between test3 and final12:09
@wwoodswhich should allow us to support upgrades 12:09
f13mether: seriously, this isn't the release to be making those sort of changes.  We have enough break the world view type of changes going on.12:09
@wwoodswhich should thus get us a *lot* more testing for test3.12:09
@wwoodskeep in mind that it's not critical, as we already have the liveCDs12:09
f13wwoods: this also assumes that we KNOW all the black magic necessary for moving you from N to N+1 by the test3 anaconda.12:09
@wwoodsand those give us a much wider testing audience12:10
f13wwoods: whereas we may need more, which would be added after test3, but if you've done the test3 upgrade, you may already be boned.12:10
f13so, Null -> Test3 -> Final probably OK.  Previous release -> Upgrade to Test3 -> Final there be dragons here12:10
@wwoodsf13: right - so, make an effort for nvr sanity and we will discuss the Arcane Magics with anaconda-team types 12:10
Lovechildspeaking of updates, what was the final outcome of the debate concerning enabling anaconda to get updates during the install?12:11
f13Lovechild: very difficult design issue around iso media and yum/rpm that prevents it for any iso based install12:11
f13Lovechild: non-iso based installs (ftp/http/nfs tree) should work fine, now that we've fixed the yum comparison bugs12:11
@wwoodshmm, that's going to be something else we'll need to test, then12:12
f13Lovechild: I'll be happy to talk to you more about it after the meeting.12:12
@wwoodsokay so: RHTS stuff! 12:12
Lovechildf13: well I'm fine with that answer really I just wanted to know if that was the plan12:12
-!- wwoods changed the topic of #fedora-qa to: Fedora QA | Meeting in progress - RHTS discussion | Agenda: http://fedoraproject.org/wiki/QA/Meetings/2007012512:12
@wwoodsf13 set us up with a trac instance for beaker (https://hosted.fedoraproject.org/projects/beaker)12:13
@wwoodsthat's where we'll be storing docs, tests, and tools that we build on top of RHTS12:13
@wwoodsRHTS itself is the basic tools from Red Hat, those come from https://testing.108.redhat.com/12:14
@wwoodsdmalcolm has submitted the package for inclusion in Extras12:14
dmalcolmas https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=22427112:14
@wwoodsfurthermore we have a xen host at http://qa.fedoraproject.org/12:14
f13we will be renaming it right?12:14
f13"Red Hat" in a package name is probably going to be not so cool12:14
@wwoods(big thanks to mmcgrath for setting that up)12:14
dmalcolmf13: what does RPM stand for?12:15
f13dmalcolm: RPM Package Manager12:15
Lovechildf13: you are right Fedora tends to take a beating for being the RHEL alpha everything we have a package called redhat something or another - it's a PR problem mostly12:15
@wwoodsRHTS Hybrid Test Suite12:15
dmalcolmIIRC I never spelled out the acronym12:15
f13Rad Hybrid TEst Suite...  (:12:16
Lovechildmuch yay!12:16
@wwoodsReally Hoopy Test Suite12:16
@wwoodsand/or System12:16
f13dmalcolm: sure, I was just pointing out past history, redhat-config -> system-config12:16
@wwoodsanyway, the rhts packages kind of give us a basic setup - dev tools, test runners, templates, etc.12:17
@wwoodswe'll build stuff on top of that base, and that will be beaker.. conceptually, it's kind of like rpm -> yum12:17
f13anaconda-runtime -> pungi   (:12:18
LovechildBen -> Jerry's12:18
Lovechildokay so how far down the RHTS road are we?12:19
@wwoodsso anyway. I think poelcat is going to be migrating the current tests from testing.108 into the hg repo at hosted.fp.o12:19
LovechildI mean can we unleash terror on a grand scale any time soon?12:19
@wwoodsand that's where we'll start putting new tests that we write specifically for fedora12:19
@wwoodsI'm going to be trying my hand at using the current tools to write a few tests over the next week12:19
@wwoodsand putting together some docs on how one grabs the tools, sets up the environment, writes a test or three, packages those tests, submits them for inclusion in beaker, etc12:20
dmalcolmwith the rhts-devel package you can run a test, and then turn it into an RPM thast can easily be installed on a test host12:20
@wwoodsand hopefully that's what we'll be doing during FUDCon12:20
@wwoodswriting tests and running them against F7t112:20
dmalcolmI hope to create a web UI for scheduling test runs on rawhide etc12:21
@wwoodsor just screwing around with Test112:21
Lovechildwow that soon, I'm stunned12:21
@wwoodsLovechild: honestly, writing a test is pretty dang easy12:21
@wwoodswith a few good examples it should be pretty easy to build up a nice library of tests12:21
Lovechildwwoods: if a moron or a slightly trained monkey can do it I'll have a go12:21
@wwoodsand then, as dmalcolm mentions, we start building tools to grab and run tests12:22
@wwoodsso, for example, the updates tool can signal the test lab, like12:22
@wwoods"hello! there's a new httpd package available. run all the httpd tests you've got."12:22
@wwoodsand it'll do so, and report the results12:22
Lovechilda puplet for testing, sounds like run12:23
@wwoodshmm, I hadn't considered that idea - a signal to the actual testers from the updates-tool12:23
@wwoodsthat's kind of cool. heh. "hey! test out this package!"12:23
Lovechildit would aid in test coverage12:23
Lovechildwould be great for the updates-testing repos12:24
-!- Oldma1 [n=Oldman@ip68-6-209-67.sd.sd.cox.net] has left #fedora-qa []12:24
@wwoodsanyway RHTS is the focus for now - we'll talk about updates-testing more once F7 is calmer12:24
@wwoodsonce it's out, we'll need to focus on updates-testing anyway12:24
f13once the post fudcon valium kicks in...12:24
@wwoods(not that we can ignore it now, but there's bigger, scarier things going on)12:25
@wwoodsvalium kicks in / hangovers wear off12:25
Lovechildnot to mention F7 testing is way more prone to go boom in entertaining ways12:25
@wwoodsyeah, FC6 is fairly stable these days12:25
@wwoodsI don't think we are in desperate need of complete tool overhaul for it right now12:26
@wwoodsso yeah, F7 is where the fun is.. but continue keeping an eye on updates-testing, naturally12:26
@wwoodsso yeah, quick recap of actions: i'm supposed to be writing some example tests and docs12:27
LovechildI'd personally stopped testing FC6 to move to F7 - I don't have enough machines or time to do testing on both currently12:27
@wwoodspoelstra will be setting up the source repo for beaker (tests and new tools)12:27
@wwoodsdmalcolm's at work getting the current rhts tools into Extras12:27
Lovechildanyways, the family is moaning things like make dinner or suffer.. I'll have to jet12:27
@wwoodsFUDCon will be test-writin' time12:27
@wwoodsand.. yeah, 90 minutes are up, and I need some lunch12:28
@wwoodsany last thoughts?12:28
dmalcolm(i) status of when fedora-qa mailing list (ii) when will next IRC meeting be?12:28
@wwoodsdmalcolm: ah! thanks12:28
@wwoodsThe discussion about fedora-qa mailing list got bogged down in discussions about whether to keep the -list suffix and/or fedora- prefixes12:29
@wwoodsI might just go to IS/IT myself and get fedora-qa created - I don't really care for the -list suffix but if someone can give me a reason to keep it, I"m all ears12:29
@wwoodsas for the next meeting.. same time, same day of the week (unless anyone objects)?12:30
@wwoodsI think the meeting after FUDCon we can do 0100UTC to accomodate the other half of the world12:30
@wwoodsor.. third12:30
@wwoodsor something.12:30
Bob-LaptopI will be on my way to boston this time next week12:30
@wwoodsBob-Laptop: hmm. maybe we should do the meeting during FUDCon then?12:31
@wwoodsor before it?12:31
* f13 gets lunch12:31
-!- f13 [i=jkeating@fedora/ender] has left #fedora-qa []12:31
Bob-Laptopwwoods: I know a lot of us are traveling on thursday, I would think a special meeting on saturday would be the bomb12:31
@wwoodswait, hell yeah, an in-person meeting (for those lucky enough to be at FUDCon) would be great12:32
Bob-Laptopwwoods: then we can fill in the others that are not there with what we are doing and maybe get some colab going12:32
@wwoodsbut we have to discuss on IRC as well, which gets tricky for those out of the room12:32
@wwoodshow about we'll say it'll be Saturday, time TBD for now12:33
Bob-LaptopSorry I was late BTW12:33
@wwoodsI guess we can just pick a time but I need to figure out what's going on Sat. first12:33
@wwoodsso yeah, I'll do that ASAP12:33
@wwoodsBob-Laptop: no big deal.. I'll put up the log in a few minutes12:34
@wwoodsokay, I'm declaring this meeting officially done12:34
@wwoodsimma grab some lunch and post notes on the wiki, but I'll still be around to talk after that12:34
--- Log closed Thu Jan 25 12:35:19 2007

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!