18:00:14 #startmeeting FESCO (2015-04-22) 18:00:14 Meeting started Wed Apr 22 18:00:14 2015 UTC. The chair is jwb. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:14 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:00:14 #meetingname fesco 18:00:14 The meeting name has been set to 'fesco' 18:00:16 #chair ajax dgilmore jwb mitr nirik paragan rishi thozza sgallagh 18:00:16 Current chairs: ajax dgilmore jwb mitr nirik paragan rishi sgallagh thozza 18:00:22 #topic init process 18:00:23 hi 18:00:27 * nirik waves 18:00:32 hi 18:00:40 * jkurik is here 18:00:57 so i think mitr and hozza will both be missing today 18:01:02 as is ajax iirc 18:01:09 hey all 18:01:29 let's wait for sgallagh_afk rishi and dgilmore 18:01:50 if we dont' get at least 5 in the next 5 min, we'll cancel... 18:03:00 * mattdm is here 18:03:49 imcleod will be a few minutes late but can talk about state of vagrant if meeting happens 18:04:40 kinda hoping it will... 18:04:48 nobody else said they were going to miss that i remember 18:07:42 so.... 18:08:49 * rishi waves 18:09:09 ok, that's 4. 18:09:23 * imcleod present 18:09:25 I was walking home from the office for the meeting. Sorry for being late. 18:09:32 dgilmore? 18:12:35 sigh 18:12:53 proposal: Defer to next week. Encourage people to discuss and vote in-tickets 18:13:22 yeah, guess so 18:13:35 oh wait. 18:13:47 dgilmore popped up on another channel, he might be around now... 18:13:50 looks like no quorum today so +1 to jwb proposal 18:14:14 * jwb waits patiently 18:14:22 Let's wait a bit more. 18:14:30 * rishi reads the F23 schedule till then 18:14:36 * paragan is here 18:14:46 paragan, yep, knew that :) 18:18:37 * jwb continues to wait for quorum 18:20:05 hey all 18:20:14 oh great. let's move on 18:20:24 #info rishi paragan nirik jwb dgilmore are here 18:20:43 #topic #1428 cloud vagrant box 18:20:44 .fesco 1428 18:20:45 jwb: #1428 (cloud vagrant box) – FESCo - https://fedorahosted.org/fesco/ticket/1428 18:20:45 https://fedorahosted.org/fesco/ticket/1428 18:20:59 ok, so a revisit on this ticket 18:21:14 imcleod: still here? 18:21:17 Yup 18:21:34 it sounds like websites should only have a minor tweak. the cloud people say this is very important. rel-eng said it wasn't huge last week 18:21:48 perhaps we should get also some qa input? 18:21:51 the Cloud sig is doing all the QA 18:21:56 and it's non-blocking 18:22:09 jwb: speaking as a bystander... I also pinged kushal to make sure someone was corralling said QA work 18:22:20 I still stand by Change owners dropped the ball, its really too late to go adding deliverables, websites and qa are impacted by it much more than releng is 18:22:36 dgilmore, er, except i just said they aren't 18:22:46 I think we set a really bad precident by letting it in 18:22:48 because websites has a small tweak to make, and Cloud SIG is doign all the QA 18:23:22 * nirik isn't sure how much qa there has been from the cloud sig (no offense) 18:23:50 * dgilmore is with nirik here. 18:23:53 nirik: I've done scratch builds of the submitted ks content and verified that they lauch successfully in the two Vagrant environments. 18:24:04 well, that's a point. it comes down to whether we value punishing people for not following process or delivering what is being explained as an important feature for cloud 18:24:25 I guess I am a weak +1, but also could we have a checkoff that someone tested them before final release? just a 'I tested RC-N and it worked' thing like we have for spins... 18:24:28 yep, any other deliverable needs more qa backing than it is now... it's not just having one more thing, it needs release criteria, reasonable tests etc. 18:24:38 right. 18:24:45 dgilmore: I've been doing a lot of the QA for cloud stuff - but only the base cloud image 18:25:02 and saying "oh, it's not release blocking" is fine, but I don't want fedora shipping broken things, it makes the entire project look bad. 18:25:10 I'm tracking all that for F23 and will make sure things are in order next time around 18:25:28 yeah, there's so many deliverables now. ;( 18:25:28 * imcleod puts his hand up to QA the vagrant non-atomic boxes - my initial release criteria is that it is possible to "vagrant ssh" into the image. 18:25:57 Which actually ends up confirming that a whole slew of things work. It boots, yum works, ssh works, and content is rsynced into the box. 18:25:57 imcleod: we can work on it and get it out to the lsits for a vote if you like, outside the meeting 18:26:03 roshi: Roger. 18:26:08 imcleod: I can also "volunteer" kushal to assist, but I believe he's interested already :-) 18:26:12 Just wanted to get the rough parameters out there. 18:26:16 * roshi sits back down :p 18:26:22 From my point of view, it seems like we have a weird split in strictness -- or in the punishment model, if you want -- for things which are image-based deliverables vs. features delivered as packages 18:27:02 mattdm: how so? 18:27:12 For things which are rpms, we can ship 0 day updates, but something like this ends up being a 6 month delay on the whole thing 18:27:30 * nirik voiced his vote, will stay quiet now. ;) 18:27:38 mattdm: that comes down to the distribution models we have chosen 18:27:41 i'm also a weak +1 18:27:49 * dgilmore is -1 18:28:00 But beyond that, for "non-blocking" features that are in packages, we (and I'm counting when I was on fesco) we tend to be like "eh, okay", even if the feature process was problematic 18:29:00 I'm also a little bit confused of the _process_ of this vote here. This isn't a new feature -- the non-vagrant images were part of the approved feature from the beginning 18:29:40 mattdm: no one knew that, and the change owners never said hey we are missing this one thing 18:29:42 Hopefully, we've made adjustments which will improve communication with releng of actual expected deliverables for future releases -- that seems to have been the main breakdown. 18:30:19 mattdm: for f23 there is going to be a list of deliverables defined before alpha 18:30:27 mattdm, but we're talking about the vagrant images... not non-vagrant 18:30:30 Right, communication wasn't great. But right now, we basically _are_ just missing this one thing. 18:30:39 jwb: sorry, non-ATOMIC 18:30:55 mattdm: in this case no one but the change owners knew anything was missing, and its not clear they paid enough attention to even know that it was missing 18:31:13 afaik we were delivering everything for teh change 18:31:26 dgilmore: What was the specific deadline that we, the change owners, missed? I'm unclear on that. 18:31:42 imcleod: I guess alpha? or tc's before? 18:31:57 "hey, why is there no non atomic vagrant images? I wanted to use those... " 18:32:02 imcleod: the alpha needs to be somewhat testable change deadline 18:32:13 and definitly the beta has to be 100% done deadline 18:32:29 the issue was not raised until after we started RC composes for beta 18:32:29 K. So we would have needed both the kickstart and the changes to the rel-eng scripts by pre-Alpha, correct? 18:32:35 nirik -- right, that "hey" happend back in March. 18:32:55 anyhow, we are +3 (with thozza in ticket) -1 from dgilmore ? 18:32:56 imcleod: that or at leasta plan saying we need to get this thing sorted 18:33:01 mattdm: oh? 18:33:20 Back in March? The ticket is 9 days old. 18:33:20 mattdm: where? I for one missed it 18:34:24 nirik at least in cloud sig list https://lists.fedoraproject.org/pipermail/cloud/2015-March/005124.html 18:34:59 mattdm, um... you're not really highlighting anything other than the breakdown in communication here 18:35:00 ah, so I guess thats were things fell down. 18:35:03 anyhow. 18:35:24 but i don't think any of this is NEW information. we already knew it fell through the cracks 18:35:49 so as i said. it's a matter of sticking to process, or begrudgingly allowing it in 18:35:52 mattdm: the hey where is this would need to be directed to Releng 18:36:03 and that hey came after we started RC composes for Beta 18:36:27 Yeah. That *should* have happened earlier; I don't dispute that. 18:36:33 paragan / rishi: votes? 18:36:37 mattdm: and even then when I said to imcleod it needs a change and we can get it in F23, he did not say oh its supposed to be part of this F22 change 18:36:51 I see that cloud SIG will be taking all the efforts for Vagrant images, considering this we will have good images shipped as F22 GA, though it looks late request but let's consider their efforts, I will +1 18:36:59 mattdm: the first someone said it was supposed to be part of the Chage was wlaters in the FESCo ticket 18:37:04 dgilmore: Actually, I stopped trying once I was told, very firmly and explicitly, that it was way too late. 18:37:15 nirik: I am having a hard time making up my mind. 18:37:33 rishi: and it's up to you... no pressure. ;) 18:37:44 imcleod: I did say it needed a change, you could have said we already had one 18:37:45 The SIG wants it badly and is volunteering to do the QA, but adding this post-Beta seems too late. 18:38:06 nirik, it's actually not up to rishi... 18:38:19 no? 18:38:26 we only have 5 members of fesco here. we need all 5 for a passing vote 18:38:38 dgilmore: Honestly, I was working the needed delivery bits. It didn't occur to me that you had not read the change when you asserted that it was missing. 18:38:38 oh, I miscounted indeed. 18:39:12 so really, it'd be up to dgilmore to change his mind and rishi to approve if we're looking at it from that perspective 18:39:13 I don't know. I could go with a +1 to lean in favour of getting things done over bureaucracy. 18:39:14 imcleod: it wasn't in the change... walters folded into another change and it wasn't there anymore if anyone looked 18:39:28 jwb: Count me in as a weak +1. 18:39:32 this is a trainwreck of communication failure. ;) 18:39:34 imcleod: as one of teh change owners the onus is on you to make sure we have all your change in. 18:39:39 ok, so now it's up to dgilmore 18:39:52 but I don't think we should just punish everyone who might use this due to all the process failures. 18:40:00 imcleod: when i said it nees a change you should have gone but wait we already have one 18:40:05 i suggest we wrap this up soon. you guys are just arguing over known issues that nobody thinks aren't issues. 18:40:05 nirik++ yes that 18:40:33 right now i have jwb nirik rishi and paragan as +1s (weak or otherwise) 18:40:42 * mattdm is 100% in favor of fixing communication problems (and we've alaready taken steps towards that) 18:40:49 and thozza in ticket. 18:41:01 thats why I miscounted. 18:41:22 nirik, oh... 18:41:29 sometimes we need to learn hard lessons 18:41:31 i forgot thozza voted in the ticket 18:41:34 * jwb goes to recount 18:42:07 nirik, hm. thozza's vote is kind of contingent on things 18:42:12 right 18:42:23 as far as i'm concerned, it still hinges on dgilmore changing his mind 18:42:45 dgilmore, set on -1? 18:43:52 Dennis, is it too late from a _functional_, pragamtic perspective? Or is it based on wanting to stick to process as a matter of principle? 18:43:58 jwb: I think so. I think it sets a bad precident if we let it in. people will push the limits further than they already do 18:44:54 mattdm: we can physically do it. But we really should not add new deliverables just for final 18:45:05 #agreed Vote from last week stands. Change remains pushed back to F23 18:45:09 * stickster wonders why strict limits would trump encouraging innovation :-( 18:45:32 we should move on. we already started late and we have a schedule to review 18:45:59 #topic #1429 Make explicit spec file License mandatory 18:45:59 .fesco 1429 18:46:00 https://fedorahosted.org/fesco/ticket/1429 18:46:00 jwb: #1429 (Make explicit spec file License mandatory) – FESCo - https://fedorahosted.org/fesco/ticket/1429 18:46:09 * dgilmore needs to leave in 35 minutes to pick up daughter from school 18:46:22 lots of -1 in the tickets 18:46:28 i'm -1 as well 18:46:36 this seems to be knee jerk over one issue 18:46:46 Yeah, -1. 18:46:50 -1 18:47:04 yep. -1 18:47:06 -1 18:47:08 stickster, imcleod I am in (just don't know what i was volunteered for) :p 18:47:10 * kushal reads back the log 18:47:19 excellent 18:47:22 it is all licensed already, the user is free to add license info to his own spec files 18:48:15 #agreed Explict mandatory spec file licenses are not required 18:48:21 moving on 18:48:29 #topic #1430 Fedora 23 schedule proposal 18:48:29 .fesco 1430 18:48:29 https://fedorahosted.org/fesco/ticket/1430 18:48:31 jwb: #1430 (Fedora 23 schedule proposal) – FESCo - https://fedorahosted.org/fesco/ticket/1430 18:48:37 schedule 18:48:47 it mostly looks good to most people that have commented 18:48:58 I am not too worried about the misalignment with the GNOME schedule. 18:49:11 kalev noted a mismatch with gnome, which does seem to be because we're pre-emptively chopping off 2 weeks to account for slips 18:49:24 i'd kind of like to not do that 18:49:25 We got used to seeing .1/.2 in Beta or the GA due to Fedora slipping. 18:49:42 rishi, but the entire point FESCo has been trying to work on is to not slip 18:50:08 so building a schedule around content with the expectation to slip is counter-productive 18:50:13 Which meant that the .0 release didn't get enough testing, so important bug-fixes landed in .1/.2. 18:50:14 we did slip for f22 beta, but it was really close. 18:50:16 jwb: well, fesco usually asked for second half of October and earlier/better and the time we did the draft with jkurik, GNOME schedule wasn't available 18:50:20 Kind of a chicken and egg thing. 18:50:36 * nirik agrees with jwb 18:50:45 well we don't have to add 2 weeks if you count in slips we probably will have this two weeks anyway 18:50:48 jreznik, right, i'm not saying it was done improperly. i'm asking if we can shift it now that we know 18:50:49 jwb: Right. That is why I am not worried about the "misalignment". 18:51:09 * dgilmore agrees with jwb also 18:51:15 Ideally the GNOME .91/.92 would go in the Beta, which is the case here. 18:51:54 rishi: well hopefully Alpha and Beta get more testing and bugs found then the .0 release is in better state 18:52:03 proposal: Shift the schedule out to match the Oct 31 target GA date 18:52:11 dgilmore: Yes, exactly. 18:52:47 jwb: to shift the schedule means to prolong the development phase ? 18:52:50 two weeks will lead to nov 03 18:53:00 jwb: that would only be one week 18:53:17 if we wanted to ship in October 18:53:20 it would technically be 1 week and 3 days 18:53:41 also count with flock 18:54:12 flock is in august, after alpha 18:54:58 as the proposed schedule stands its the best tie in with flock 18:55:03 now, if you add one week, it means last week of alpha will be during flock if I look properly 18:55:10 for me makes sense to slip for two weeks or leave it as it is 18:55:19 we have alpha out and a quieter week 18:55:19 jkurik: yep 18:55:30 jreznik, only if you add the week during the development phase. why can't you add it between alpha and beta? 18:55:49 because it's usually beta that slips and causes problems anyway 18:56:13 it's not about time between milestones but when we're able to get reasonable working first TC 18:56:39 but adding more time between milestones is possible, shortening is path to ... 18:56:47 if someone would like to make a counter proposal to approve the schedule as-is, please do so 18:57:29 actually, i have an alternative proposal 18:57:31 well, kalev asked for two weeks, what we're going to fix with one week? 18:57:54 jreznik: I don't think we really need 2 more weeks for GNOME. 18:58:07 one week and 3 days. and we're fixing it to match our rule of may 1/oct 31 18:58:15 i.e. we're getting back to a time based release 18:58:30 I am in favour of the current GA of 20th October. 18:58:32 not a release where we kind of are time based but oh maybe we account for slips *wink wink* 18:58:42 oct 31st is a saturday tho. 18:58:46 i don't care 18:58:49 jwb: Agreed. 18:59:58 frankly, i'd like the Fedora Program Manager to just TELL us what the damn schedule is. and then have FESCo base content off of that 19:00:21 because we do this song and dance every release and it's getting very old 19:00:54 well, we could say: The tuesday nearest May 1st/Oct 31st 19:00:57 we have an offical FPM role on the Council now. let's just let them do their job and we'll do what we do, which is drive the content of the distro based on the boundaries of the schedule 19:00:58 and always work back from that 19:01:28 jwb: what's what I'm doing - usually fesco wanted october release, a bit earlier... 19:01:48 if we want to strictly follow the rule and october - it's clear, oct 27 19:01:51 jreznik, no, what you are doing isn't telling us. you're asking us for approval. i don't want FESCo to be able to approve or disapprove it 19:02:08 i want you to talk to all the other parties involved and set the schedule and let us know. 19:02:19 which is what a Program Manager does :) 19:02:25 proposal, accept schedule as is. 19:02:32 dgilmore: +1 19:02:41 im +1 to it 19:02:48 we told you what we thing is best, program manager just proposes, does not have decision power (at least in red hat ;-) 19:03:01 what a shame. 19:03:40 jwb: So this is the last item on the agenda, right? 19:03:50 aside from open floor and next week's chair 19:03:56 * rishi is getting really hungry 19:04:00 * nirik doesn't understand why jwb is getting mad at jreznik following the process WE SETUP. If you want to change that fine, but no need to get upset about it 19:04:24 I'd prefer an extra week or two in there personally. 19:04:55 nirik, i'm not mad at jreznik at all. not in the slightest. i'm mad at us for continually being really poor at doing something i don't think we're even really entitled to do, which is set the schedule 19:05:14 I don't have any problem with early nov neither and two weeks should work - just I know how fesco reacts to any schedule that lands to november 19:05:16 alright. 19:05:25 I am +1 to the current schedule and OK with pushing the GA back to Oct 26 or Nov 3, but not more than that. 19:05:37 And I am not OK with pushing the schedule back just because of GNOME. 19:05:53 i am with rishi there 19:06:08 And with that, I am going to run and grab some food. 19:06:13 thanks rishi 19:06:13 proposal: add 1 week after alpha ships, before beta freeze, move later milestones out a week and release oct 26th 19:06:19 we can always try to aim the dates but there's always something like flock, gnome schedule, other schedules and that just makes sense to talk about it as more eyes sees more :) 19:06:26 nirik, +1 19:06:36 nirik, +1 19:06:37 nirik: that is fine also 19:06:38 that gives us one more week of beta TC's... 19:06:47 ok, that's quorum 19:06:48 nirik: +1 19:06:49 which I think really helped doing so many 19:06:54 I think the current alpha date is the best possible alpha date 19:07:03 nirik: I'm not happy about changing the cadence between milestones but we did it last year too to skip flock 19:07:13 #agreed add 1 week after alpha ships, before beta freeze, move later milestones out a week and release oct 26th 19:07:16 Oct 26th is my b'day! Can't be better than that. 19:07:25 * rishi -> AFK 19:07:29 jreznik: yeah, beta seems to be the one we have most problems with, but yeah 19:07:29 moving on to open floor 19:07:35 nirik: yes, early TCs are making the real difference - on the other hand it has costs on dgilmore's and QA sanity :) 19:07:53 #topic Open Floor 19:07:54 jwb: we made the decision on the schedule in approx. 20 minutes - it is not so bad :-) 19:08:01 jkurik, :) 19:08:04 rishi: it's oct 27, sorry :) 19:08:25 jreznik: I am hoping to make changes to the rawhide process before branching that makes nightly composes look like a release compose not a subset of it 19:08:49 TC's every night! 19:08:54 jreznik: which will hopefully mean that we can do less TC's and more frequent testing 19:09:00 nirik: pretty much :) 19:09:10 that's awesome 19:09:18 that would be nice 19:09:24 is that notable for open floor? 19:09:46 jwb: if you wanted to 19:09:54 nothing for today but with adamw we talked about the idea having just snaps instead of alpha, beta pre-GA (just a bit) 19:10:08 so maybe it could fit together pretty well 19:10:14 #info dgilmore looking into making rawhide composes more like release composes to help with testing 19:10:41 anything else for open floor? 19:11:07 jwb: not here 19:11:21 * jreznik is good 19:11:30 * jkurik is fine 19:11:38 i'll chair next week again 19:11:39 jwb: it has been awhile since I ran a FESCo meeting, I can do next week 19:11:44 ok, you win 19:11:44 or you can 19:11:47 nope you 19:11:51 okay 19:11:53 :) 19:11:53 #info dgilmore to chair next week 19:12:12 anything else from anyone? if not, i'm going to close in 19:12:47 #endmeeting