18:00:49 #startmeeting FESCO (2015-07-15) 18:00:49 Meeting started Wed Jul 15 18:00:49 2015 UTC. The chair is dgilmore. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:00:49 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:00:49 #meetingname fesco 18:00:49 The meeting name has been set to 'fesco' 18:00:49 #chair ajax dgilmore hguemar jwb nirik paragan rishi thozza sgallagh 18:00:50 Current chairs: ajax dgilmore hguemar jwb nirik paragan rishi sgallagh thozza 18:00:50 #topic init process 18:00:57 hello 18:01:02 Hi 18:01:14 hello 18:01:23 hi 18:01:25 morning 18:01:59 sorry for the late agenda 18:02:14 Hello 18:02:17 I forgot about branching being yesterday when I said i would run the meeting 18:02:26 np 18:02:32 no problem, you had enough work with branching 18:02:53 #topic #1247 List of release blocking deliverables 18:02:53 .fesco 1247 18:02:54 https://fedorahosted.org/fesco/ticket/1247 18:02:55 dgilmore: #1247 (F21 System Wide Change: Remove python-setuptools-devel - https://fedoraproject.org/wiki/Changes/Remove_Python-setuptools-devel) – FESCo - https://fedorahosted.org/fesco/ticket/1247 18:02:59 lets get moving 18:03:08 re 18:03:24 so FESCo is supposed to provide this week a list of all the things that are supposed to be delivered in f23 18:03:37 .fesco 1427 18:03:38 paragan: #1427 (List of release blocking deliverables) – FESCo - https://fedorahosted.org/fesco/ticket/1427 18:04:11 gahh stupid typo 18:04:15 fun. 18:04:15 thanks paragan 18:04:22 np 18:04:48 #topic #1427 List of release blocking deliverables 18:04:51 so, how do we want to do this? 18:05:06 in the ticket I guess 18:05:07 try and brainstorm now? have some folks come up with a proposed list in ticket later? 18:05:28 based on the ticket I thought jaroslav was going to produce a list that we could sign off on 18:06:13 we should go through changes we approved 18:06:16 yes 18:06:18 to me it feels like f23 is catching quite a few people off guard. 18:06:27 perhaps this wasn't handed off to the new program manager? 18:06:35 as in they were not expecting it yet 18:07:01 can we do it as part of alpha review? 18:07:20 dgilmore: time flies when you are having fun. :) 18:07:36 thozza: maybe. the reason that we had it now was to make sure releng was able to deliver everything 18:07:42 nirik: indeed 18:07:53 right 18:08:10 thozza: if we wait a couple of weeks we may not be able to do so 18:08:27 so do a check next week 18:08:39 but I guess we can say it will not be in Alpha we will get it in before Beta 18:08:40 we need the list before alpha freeze... on the 28th? 18:08:54 nirik: that would be the latest 18:09:02 so ideally next week, as thats the last meeting before then 18:09:08 ticket says we need to decide before alpha freeze 18:09:10 yep 18:09:11 sounds good 18:09:22 * nirik added jkurik to the ticket 18:09:49 proposal work with program management to get a list of all deliverables for next weeks meeting 18:10:17 +1 18:10:24 +1 18:10:34 +1 18:10:40 +1 18:10:52 +1 18:11:09 +1 18:11:23 #accepted work with program management to get a list of all deliverables for next weeks meeting (6,0,0) 18:11:36 #topic #1461 F23 System Wide Change: Layered Docker Image Build Service 18:11:38 +1 (sorry) 18:11:39 .fesco 1461 18:11:40 dgilmore: #1461 (F23 System Wide Change: Layered Docker Image Build Service) – FESCo - https://fedorahosted.org/fesco/ticket/1461 18:11:42 https://fedorahosted.org/fesco/ticket/1461 18:11:56 sure, +1 18:12:03 +1 18:12:13 +1 18:12:20 I am +1 but i think it will be Beta before we have it in place 18:12:21 +1 18:12:38 yeah, it's going to be a lot of work and blazing somewhat new trails. 18:13:29 delaying it another 6 months would be bad, and as long as it's here for GA, it won't disturb existing users 18:14:41 indeed 18:14:42 I also think we should try to have it in place for F23 18:14:56 +1 18:15:40 i count 6 +1 18:17:53 #agreed F23 System Wide Change: Layered Docker Image Build Service approved (6,0,0) 18:18:11 #topic #1462 RPM Weak Dependencies and the install media compose process 18:18:14 .fesco 1462 18:18:15 dgilmore: #1462 (RPM Weak Dependencies and the install media compose process) – FESCo - https://fedorahosted.org/fesco/ticket/1462 18:18:17 https://fedorahosted.org/fesco/ticket/1462 18:18:59 the ticket was created late today 18:19:06 I covered my thoughts on the mailing list 18:19:24 Yes, but only as a summary of the conversation on the mailing list 18:19:45 dgilmore: I *think* my summary agrees with you, yes? 18:19:52 I'm fine with option 1 seems fine. 18:20:17 yes, that is right 18:20:18 I guess it means tooling changes. 18:20:20 sgallagh: not really 18:21:00 sgallagh: i think that anaconda needs a switch to be able to toggle installing suggets 18:21:14 but in f23 I think thats all we can do 18:21:57 right now the compose tools all use yum right? or does pungi4 use dnf? 18:22:12 nirik: they all use yum 18:22:35 which will ignore recommends right? 18:22:52 I am honestly not sure 18:23:02 nirik: it should 18:23:27 nirik: Right, which is potentially a problem as packages are starting to take advantage of Recommends: 18:23:35 (To shrink their minimal footprint) 18:24:12 sgallagh: I'm ok with the proposal 18:24:36 well, the compose tools should switch to dnf, but thats going to be a lot of work, no waving a magic 'wish this was all dnf' 18:25:01 thats a lot of work thats not even started yet 18:25:54 and it will help with the move with python3 ;-) 18:25:57 Well, we basically have two choices as I see it: 18:25:58 1) Fix our compose tools to match the package manager all of our deployed users have 18:25:58 2) Ask the FPC to disallow use of weak dependencies until 1) happens 18:26:00 so, I think for this cycle we may have to live with the yum behavior 18:26:03 it is somewhat unfortunate that the on-list discussion didn't bring this up at all 18:26:42 then, let's make it a goal for F24 (switching compose tools to dnf) 18:26:54 sgallagh: 3) tell everyone the composes don''t care about those and fix any cases in packages we need to care about. 18:27:01 Or 3) require micromanagement of comps/spins-kickstart to work around the lack of proper support. 18:27:21 well, or just change packages that matter back to requires or whatever. 18:27:22 sgallagh: what this will mean in practice is that network installs are bigger than dvd installs 18:27:42 nirik: I don't think that's realistic. The benefits of the weak deps are strongest in the important packages with big depchains 18:27:42 since dvd installs will not have reccomends available 18:28:22 sgallagh: well, its been possible for 22 releases. ;) 18:28:41 My point being that if we're advising to change the packages that matter, that's rapidly approaching the same as 2) above 18:28:57 I think that compose does not need to stick with recommends even if we were using dnf 18:29:12 number80: I don't follow 18:29:17 I suppose, but I couldn't even get fesco to forbid them before there were any guidelines in existance (ie, some were used in f22 and before) 18:29:53 sgallagh: composes should not require Recommends:, and keep the same behaviour as yum in the long term 18:30:04 number80: so your proposal would be to keep reccomends off of installs? 18:30:08 number80: I disagree completely 18:30:27 dgilmore: yes 18:30:29 The whole point of Recommends is that the default experience should be more complete than the minimal experience 18:30:48 sgallagh: we don't know yet the impact on media size on the long run 18:30:48 The installer is pretty much the definition of the "default experience" 18:30:53 number80: the issue there is when doing network installs dnf's default behaviour will install them 18:31:21 so if you do a dvd install and a network install you will get two different installs for the same options 18:31:33 nirik, dgilmore: I presume there's more to the tool switchover than s/yum/dnf/ ? 18:31:36 yes, but I would prefer having more feedback before 18:31:56 I'm not against the switch, I just say we need more data 18:32:03 sgallagh: much much more 18:32:12 yes, I am sure its a lot more. 18:32:53 sgallagh: the switch to dnf is something we want to get done in the f24 timeframe 18:33:09 we have had other thinsg we have been working on for f23 18:34:02 i don't think the discrepency between DVD and network installs is an issue 18:34:22 jwb: That ultimately depends on what ends up switching from Requires: to Recommends: 18:34:33 dgilmore: ideally combining it with python 3 18:34:36 not really. it has always been different after ~1 week anyway, because of updates 18:34:45 If it happens en-masse, we could end up with DVD installs that are just above worthless 18:34:47 there's already a discrepancy between the packages available from network and from networkless install 18:34:48 thozza: thats likely f25 at the earliest 18:35:09 dvd installs at this point are pretty worthless. at least in the traditional sense 18:35:12 jwb: updates won't pull in Recommends 18:35:39 livecds use yum also 18:35:49 so all the livecds will not have reccomends 18:36:01 unless we manage to get livemedia-creator in place 18:36:02 sgallagh, i'm aware it won't. my point was there is a size discrepency already TODAY between network and dvd 18:36:19 so i don't think it matters if there's a discrepency with f23 either 18:36:20 jwb: It's not the size that bothers me 18:36:21 could we make it an Anaconda option? 18:36:30 It's the potential loss of *functionality* 18:36:32 * nirik wonders how many things on the media will change from requires->recommends. is this a big thing? 18:36:34 number80: No 18:36:47 sgallagh, loss in DVD installs, yes? *yawn* 18:36:56 sgallagh: then how could we handle the case of recommends inflation? 18:37:02 jwb: You mean that thing we hand out at conferences? 18:37:20 sgallagh: we hand out livecds 18:37:23 number80: The only way I can think of is to just forbid Recommends: until F24 18:37:28 sgallagh, yeah. that thing that we hand out at conferences and then people install and then are faced with 500MB of updates IMMEDIATELY AFTER 18:37:29 when the tools are there. 18:37:39 currently, there's no loss as it's the current behaviour 18:37:59 * nirik notes there's only the server dvd now 18:38:10 number80: The current behavior is predicated on there not being any weak deps in the repositories 18:38:12 sgallagh, the useful shelf life of DVD media is literally less than a month 18:38:20 but I'd like to see the same compose with both options and see the impact before deciding 18:38:31 beyond that, it's obsolete and everything will be updated after the install anyway 18:38:42 number80: with what options? 18:39:04 dgilmore: one with recommends enabled and the other recommends disabled 18:39:06 jwb: I don't see why you continue to talk about the updates. Updates will only update the packages that are already installed. 18:39:13 number80: we can not do that 18:39:16 The packages already installed will be a reduced set 18:39:21 number80: it is just not possible 18:39:24 Likely with reduced functionality. 18:39:32 dgilmore: even locally? 18:39:34 That will continue forward without manual intervention by the user. 18:39:37 sgallagh, because you continue to miss my point. DVD installs are NOT useful, nor are they a major concern 18:39:47 * dgilmore does note that if they enable updates at install time the point is moot 18:39:48 sgallagh: once that one dnf bug is fixed yes... ;) 18:40:04 it is purely disconnected dvd installs effected 18:40:10 well, I suggest that we move this discussion on the list, especially as it won't affect F23 18:40:15 right now dnf reinstalls recommends on upgrades 18:40:21 jwb: The same is true of the livecd installs except insofar as they are more meticulously curated. 18:40:30 correct. 18:40:33 nirik: OK, yeah that's unexpected. 18:40:49 number80: It *does* affect F23. 18:40:52 yeah, and undesired I think. 18:40:53 sgallagh: unexpected behaviour should be expected with dnf 18:40:57 the only value live media brings is less time spent getting into the real install, or trying things out before install 18:41:06 dgilmore: s/unexpected/undesired/ 18:41:14 sgallagh: composes tool will be using yum for F23 so it won't affect F23 18:41:21 or I'm missing something 18:41:33 number80: You are missing something 18:41:37 ? 18:41:47 https://bugzilla.redhat.com/show_bug.cgi?id=1229701 18:41:49 Packagers are making changes to their packages that will result in yum installing fewer of their deps 18:41:54 number80: there will be inconsitent behaviour in f23 depending on how you install 18:42:23 any proposal? or moving to next week? 18:42:26 number80: if a package changes from Requires: foo to Recommends: foo, the install tools won't pull in foo... 18:42:31 For example, with yum, you could only do Requires:, so if your package was most useful with a particular plugin, you'd have to Requires: it 18:42:49 With the weak deps, you can now Recommends: it so it's possible for people to remove it if they truly don't want it there 18:42:56 dgilmore: yes, but we can't fix that except forbidding the use of Recommends 18:43:18 number80: That's what I'm getting at (and have been saying since the beginning) 18:43:29 We either fix the tools to support weak deps or forbid them until we can. 18:43:49 Yes, but I'm willing to have that small discrepancy as we need data to build consistent guidelines around weak deps 18:43:49 I'm not sure the changes are widespread enough to matter, but I guess I could support asking for people not to use them in f23. 18:43:50 Allowing them without fixed tools is going to cause headaches and user confusion. 18:43:53 they are already used anyway 18:44:20 sgallagh: or we accept the inconsitencies 18:44:38 dgilmore: Right; I'm not sure I'm prepared to do that 18:45:09 apologies, didn't see a meeting announcement in -devel 18:45:42 there was none 18:45:47 ajax: sorry it was late 18:46:00 thozza: there was 18:46:06 so, discuss on list for another week? try and vote on one of the two possible actions? 18:46:16 nirik: yeah I think so 18:46:38 which? :) 18:46:50 dgilmore: it was after I left work, so I didn't see it 18:46:52 Just to throw it out there: 18:46:52 Proposal: Disallow weak dependencies on any package on the various install media of Fedora 23 18:47:23 -1 18:47:24 -1 18:47:56 sgallagh: it is not enforceable 18:48:07 I agree 18:48:13 -1 18:49:35 -1 18:49:37 I can try a grep of whats using them currently to send to the list if that data would be of help 18:49:46 nirik: Yeah, please do 18:49:54 any more votes? 18:50:10 0 here (want to look at it more) 18:50:11 just today I added suggests to a package 18:50:13 0 18:50:30 at best, we can advise to wait F24 18:50:31 so I think it is a bad idea to ban them 18:51:02 #agreed no consensus: Disallow weak dependencies on any package on the various install media of Fedora 23 (0,2,4) 18:51:17 so lets discuss further on devel list 18:51:57 #topic #1460 Worfkflow/guidelines for bugs requesting change to defaul preset policy 18:52:00 .fesco 1460 18:52:02 dgilmore: #1460 (Worfkflow/guidelines for bugs requesting change to defaul preset policy) – FESCo - https://fedorahosted.org/fesco/ticket/1460 18:52:02 https://fedorahosted.org/fesco/ticket/1460 18:52:15 so i think this went a bit far to not reach out to me 18:52:32 so what about a special package? 18:52:32 it was an unfortunate series of bad timings and using bugzilla 18:52:46 more people watching the bugs and helping are welcome 18:52:48 policy 18:52:49 .fesco 1460 18:52:50 dgilmore: #1460 (Worfkflow/guidelines for bugs requesting change to defaul preset policy) – FESCo - https://fedorahosted.org/fesco/ticket/1460 18:52:53 gahh 18:53:05 I get around 200-300 bug emails a day 18:53:13 so its easy for me to miss bugs 18:53:22 I've already said I will take a look this week. :) 18:53:30 dgilmore you are too busy. I think we need more people 18:53:32 hopefully we can get any backlog handled. 18:53:41 ideally people just send pull requests on pagure for patches 18:53:43 dgilmore: As I suggested in the bug, I think it would be prudent to get the WG liaisons on the watchlist 18:54:07 dgilmore: That's a new process (and one that will make this easier, sure) 18:54:09 sgallagh: I am not opposed 18:54:14 sure. 18:54:15 But one that people haven't had time to get used to yet 18:54:16 more eyes are welcome 18:54:45 sgallagh: sure, it is new 18:55:05 bugzilla is a really bad forum for communications :( 18:55:09 which is kinda sad 18:55:24 sgallagh: I did add nirik to the watch list 18:55:34 I just added myself as well 18:55:36 anyone is free to add themselves to it 18:55:50 pkgdb allows it 18:55:58 what about special package and special mailing list as the assignee 18:55:59 ? 18:56:09 seems like overkill for a single file. 18:56:10 If people can think of a much better workflow I am all ears 18:56:28 thozza: I do not know that would work well 18:56:54 at least it would reach more people 18:57:06 who can fix things 18:57:41 but yes, it's kind of overkill 18:57:42 why wouldn't adding those people to the package do that too? 18:58:06 * dgilmore notes that very few people can actually build fedora-release 18:58:13 it would I guess. it was just an idea 18:58:49 anyhow, shall we move on then? or ? 18:58:51 it is in the secure-boot channel same as the kernel as pjones plans to do something so that the hardware can verify things 18:58:58 that's another thing, that the Bush are now "just" in post 18:59:03 bugs 18:59:11 sorry for typo 18:59:31 dgilmore: but I'm not actually sure when we're ever going to do that, so if you want to move it back, that's fine. 18:59:50 pjones: maybe we shoudl take it back out 18:59:54 okay by me. 19:00:04 we can always add it back 19:00:25 yeah; for now do whatever makes things most convenient for you. 19:00:33 do you want to put it into systemd again? 19:00:46 thozza: systemd guys did not want it there 19:00:56 thats why it was moved 19:01:09 so where? 19:01:09 yes 19:01:18 they had been asking for it to be taken out of systemd for years 19:01:23 fedora-release still makes the most sense, to me 19:01:23 that's why I'm asking 19:01:32 me too. 19:01:39 dgilmore I know 19:01:42 thozza: I do not think it needs to move 19:01:50 I don't want to put it there 19:02:08 so I misunderstood you I guess 19:02:09 I think that having some more eyes on things may help things not fall through the cracks 19:02:29 yes, that's true 19:02:34 thozza: in this case it was a bunch of missteps that resulted in the bug being missed 19:02:46 some unfortunate timings 19:03:12 and the fact that I get so much bug email that it is too high in noise for me 19:03:43 thozza: had they pinged on the bug, in email or irc it would have gotten attention and been dealt with 19:03:56 but they waited a month and then filed the fesco ticket 19:04:39 I asked Michal to file the ticket so we can discuss the thing and possibly help you 19:04:55 it was not meant as complaining 19:05:08 thozza: I did find it a bit rude to file the ticket without pinging me once before. 19:05:09 don't take it wrong 19:05:27 thozza: but I am not the sole person that can do things 19:05:40 it was more about the process 19:05:45 there is a few people that are watching bugs and can commit to the upstream repo and do builds 19:06:17 thozza: the best way to get changes in is to do as sgallagh did and file a pull request in pagure 19:06:30 thozza: I am not teh only one that can merge them in 19:06:45 maybe we should document that better and communicate it 19:07:05 so let's state that and we are finished 19:07:12 okay 19:07:18 sounds good to me 19:07:22 the info on the wiki might use some updating there too (according to the mailing list) 19:07:33 nirik: I am sure 19:07:50 I know we have not updated all the docs with teh move to pagure 19:08:47 proposal update the docs to have people send pull requests via pagure and coummicate that as much as possible 19:09:06 sure. +1 19:09:43 +1 19:09:45 +1 19:09:58 +1 19:09:59 +1 19:10:47 +1 19:11:10 +1 19:11:44 #accepted update the docs to have people send pull requests via pagure and coummicate that as much as possible (7,0,0) 19:12:22 #topic Next week's chair 19:12:32 who wants to run the meeting next week? 19:13:17 * number80 will be at europython next week 19:13:35 I haven't done it in a while 19:13:53 #info sgallagh to run next weeks meeting 19:13:55 thanbsk sgallagh 19:13:59 thanks even 19:14:08 #topic Open Floor 19:14:10 thanks 19:14:40 does anyone have anything? 19:14:42 FYI, I am continuing to try and work on the passphrase policy. Hopefully I will have a draft to propose before too long. 19:14:45 no 19:14:51 nirik: cool 19:15:16 great 19:15:25 Excellent 19:15:34 #info nirik continues to try and work on the passphrase policy 19:16:32 #endmeeting