14:00:33 #startmeeting Workstation WG 14:00:34 Meeting started Mon Jan 15 14:00:33 2018 UTC. The chair is stickster. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:34 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:00:34 The meeting name has been set to 'workstation_wg' 14:00:37 #meetingname workstation 14:00:37 The meeting name has been set to 'workstation' 14:00:40 #topic Roll call 14:00:43 .hello pfrields 14:00:52 stickster: pfrields 'Paul W. Frields' 14:01:01 .hello cschalle 14:01:02 cschalle: Sorry, but you don't exist 14:01:47 .hello catanzaro 14:01:47 mcatanzaro: catanzaro 'Michael Catanzaro' 14:02:37 .hello mclasen 14:02:38 mclasen: mclasen 'Matthias Clasen' 14:03:01 #chair cschalle mclasen mcatanzaro 14:03:01 Current chairs: cschalle mcatanzaro mclasen stickster 14:03:08 * stickster looks for rest of the crew 14:03:19 .hello petersen 14:03:19 juhp: petersen 'Jens Petersen' 14:03:30 So RH America works on MLK day? Or everyone just really devoted to showing up on a holiday? 14:04:08 mcatanzaro: we do work on MLK day 14:04:12 we always work 14:04:34 #chair juhp 14:04:34 Current chairs: cschalle juhp mcatanzaro mclasen stickster 14:04:50 Interesting... MLK day is huge here in Missouri. But you get silly holidays like Columbus Day off... OK. 14:04:53 We have quorum, let's go ahead but hopefully we'll see one or two other folks show up 14:05:03 mcatanzaro: We don't get Columbus day either ;-) 14:05:04 "Huge" meaning "at least I don't have to go to work today" 14:05:05 .hello otaylor 14:05:06 otaylor: otaylor 'Owen Taylor' 14:05:11 .hello kalev 14:05:12 kalev: kalev 'Kalev Lember' 14:05:18 #chair otaylor kalev 14:05:18 Current chairs: cschalle juhp kalev mcatanzaro mclasen otaylor stickster 14:05:46 also I am nin 14:05:51 Norway atm, so 14:05:57 MLK doesn't happen here :) 14:06:06 So I'm going to inject a topic first 14:06:12 before we get to the agenda 14:06:20 #topic Workstation WG composition check 14:06:56 #info (1) stickster looking to pass secretarial baton, after a couple years of running this meeting 14:07:30 #info (2) we should also check membership rolls; for instance rdieter doesn't seem to be on IRC since his job change and we should check his interest 14:08:02 We don't need to settle these items here, I just wanted to bring them up and we can handle via the list 14:08:03 we could maybe try to have a rotating chair, so that it's not one single person who has to run all the meetings 14:08:44 kalev: true, FESCo does that and it seems to work OK 14:08:45 kalev: yes that's not a bad idea 14:09:25 Unless somebody wants to volunteer to chair them all, rotating chair seems like a good solution. 14:09:26 rotating seems ok to me too 14:10:06 Yeah - I don't think I could handle alll, but I could handle a few (at least with a meetbot quick reference sheet) 14:10:37 If not one's opposed, I'll set up an alphabetical rotation. We have https://fedoraproject.org/wiki/Zodbot for reference, otaylor 14:11:17 stickster: For the record, FESCo's usual approach is to just ask at the end of the meeting who will be around to run it the next time. 14:11:17 The best reference sheet is the last meeting's log ;) 14:11:28 It's usually just on the honor system that everyone takes a turn 14:12:06 sgallagh: ah, interesting. That sounds just as workable if people do the expected 14:12:26 Oh gosh, then we have to think and volunteer sgallagh... alphabetical by default sounds easier to me. We can just trade spots whenever we have time conflicts. 14:12:47 Let's spend a long time arguing about this minor procedural point! 14:12:48 I think alphabetical migth work bettter here - mcantazaro seems to get too much of the "volunteer" load 14:12:54 yeah mcatanzaro, exactly 14:13:00 Thanks otaylor :D 14:13:10 #agreed set up a rotation starting at the next meeting 14:13:20 * sgallagh avoids asking "alphabetical by what name? IRC nick?" :) 14:13:24 #action stickster email the list with proposal/details for chair rotation 14:13:29 sgallagh-- 14:13:33 I'm taking away a cooki!!! 14:13:37 * sgallagh sadface 14:13:39 lol 14:13:57 but seriously sgallagh++ thanks for the info :-) 14:13:57 stickster: Karma for sgallagh changed to 11 (for the f27 release cycle): https://badges.fedoraproject.org/tags/cookie/any 14:14:07 sgallagh: middle name ;o) 14:14:38 * mclasen notes that he'll be travelling for the next 3 mondays 14:14:48 so skip the 'c' for the first round... 14:15:10 * otaylor will also be gone for the next meeting 14:15:16 * stickster too 14:15:20 * kalev too 14:15:22 #chair rdieter 14:15:22 Current chairs: cschalle juhp kalev mcatanzaro mclasen otaylor rdieter stickster 14:15:27 i suspect we'll skip a meeting ;-) 14:15:36 Hey, rdieter is here! cool 14:15:42 or we can do an in-person one in brno 14:15:44 hola 14:15:49 might come close to quorum there 14:15:51 mclasen: yeah 14:15:56 OK, moving on to agenda 14:16:18 #info Agenda is in the group's Pagure issues list 14:16:24 #link https://pagure.io/fedora-workstation/issues?status=Open&tags=meeting 14:16:36 #topic Reduce initial setup redundancy 14:16:38 #link https://pagure.io/fedora-workstation/issue/21 14:16:57 OK, this looks a bit contentious :-( ... mcatanzaro can you fill us in on the current facts? 14:17:14 Yeah... so, FESCo has rejected our change proposal, and asked us to modify it to remove stuff from g-i-s instead of from Anaconda 14:17:22 Which none of us really want to do 14:17:39 I think we should just go back to fesco with that and explain our position better. 14:17:59 So the high-level options here are (a) modify the proposal as requested by FESCo, or (b) withdraw the proposal. Or yeah, (c) try again with FESCo. 14:18:08 Yeah, I'm not understanding why this went completely 180 degrees around 14:18:21 It's interesting because last time FESCo approved the proposal, yes, and I believe that was unanimous or close to it 14:18:26 mcatanzaro: was taking this to fesco a request from the anaconda team? Not that we can bypass fesco now, but this actually seems like something that working groups are supposed to handle - the user experience of an edition 14:18:30 But this time it was unanimously rejected. I believe kalev missed that meeting. ;) 14:18:30 I can try to represent FESCo's stance 14:18:48 nobody from the Workstation side was there to talk about the change when fesco was discussing that 14:19:30 sgallagh: sure, would be appreciated 14:19:58 FESCo likes the idea of not duplicating this work in both Anaconda and G-I-S, but the problem that we saw was that the approach being taken would require all of the other desktops to implement their own solution. 14:20:12 Because they could no longer rely on getting those answers from Anaconda. 14:20:22 This set too high a bar for other desktops. 14:20:27 I don't think that is true ? 14:20:31 sgallagh: I don't think we were proposing that those capabilities be turned off entirely - just for the workstation installer 14:20:32 I don't think that is the case. As I understand the proposal, other desktops would just keep on using existing anaconda code 14:20:34 Now, if that interpretation is wrong, I'm sure that we can reopen the conversation 14:20:35 otaylor: It's a major systemwide change proposal, so I suppose I figured going through the normal change process was appropriate. 14:20:46 it's just workstation that would switch to not showing certain anaconda screens 14:21:16 kalev: that's how I understood it too 14:21:18 sgallagh: I definitely don't think we'd want to do that for server 14:21:20 kalev: Well, don't forget to consider the case of the generic netinstall 14:21:33 sgallagh: doesn't that use Anaconda too, though? 14:21:42 You can't necessarily control which spokes will be visible there 14:21:42 Yeah, the proposal is to not change other desktops at all. I read the FESCo meeting log and I didn't get that sense of the decision, it looked to me like FESCo just didn't want two different ways of installing Fedora depending on which edition you chose. 14:22:03 all we have been asking for is to make anaconda flexible enough that we can disable the bits that are not needed in the workstation case 14:22:15 So just for Workstation Live? 14:22:27 Regarding netninstall... there was some controversy over this too. adamw wants Workstation netinstall to match the live netinstall behavior; my proposal was to leave netinstall unchanged instead. 14:22:29 sgallagh: oh, I see -- IOW installing GNOME via netinstall. But isn't that a vanishingly small # of users, and also do we want to gear experience toward that small a sliver of the pie? 14:22:45 * stickster stops splitting discussion and listens 14:22:52 stickster: I'm pretty sure it's the primary way people do mass-deployments 14:22:58 Since you can't really kickstart the live media 14:22:59 Andy mclasen, anaconda is already sufficiently-flexible... there is a configuration file we can use to disable the spokes. I've tested it and it is (mostly) working. 14:23:02 *And 14:23:20 sgallagh: mass deployment seems like a different case - I think what we really care about is the reviewer/download case is slick and polished and not confusing 14:24:02 mcatanzaro: that makes the fesco decree all the more galling 14:24:08 and in the mass deployment case, they'd see whatever Anaconda is doing currently + normally 14:24:12 Anyway, our WG was created by and is clearly subservient to FESCo, so either we convince FESCo to change its mind, or we sadly don't do this. 14:24:39 it feels to me like this decision doesn't seem to align well with the reason we created the WGs 14:24:50 mclasen: Instead of hostile language, can you please be satisfied that I'm here trying to listen? 14:25:00 esp. since the change in this case is designed not to impact other editions, etc. 14:25:26 oh, I appreciate that you are here, but you get to listen to me being pissed off by this decision... 14:25:26 stickster: That didn't come across to me in the original proposal, FWIW. 14:25:28 bringing it back to Fesco for reconsideration seems reasonable 14:26:05 So the next FESCo meeting is Friday 2018-01-19 at 1700 UTC / 11:00am US-Eastern IIRC 14:26:11 In my *personal* case, it was because I wasn't terribly comfortable with having different install experiences depending on which ISO you downloaded. 14:26:28 sgallagh's explanation for fesco's motivations satisfies me, and think it's wise to assume good intentions too. So, probably best to go back and clarify this is for workstation only and that other desktops should be unaffected 14:26:32 sgallagh: I agree it is not ideal 14:26:44 QA wasn't thrilled with the idea either, since it increases the set of paths they need to cover 14:27:06 sgallagh: I don't agree with that... it *definitely* reduces the set of paths needed to cover. 14:27:13 yet we can't get ourselves to drop any installation methods... 14:27:36 sgallagh, I understand where you are coming from, but wasn't part of the motivation for the editions to move away from the lowest common demoninator style development for Fedora? 14:27:39 It would be unfortunate that anaconda implemented the ability to disable spokes, and then no one be allowed to use that feature 14:27:43 Because currently you have to test workstation installation separately for the cases (a) created root password and initial user in anaconda, (b) created root password but not initial user in anaconda, (c) created initial user but not root password in anaconda 14:28:15 mcatanzaro: ah good point 14:28:26 mcatanzaro: But they still need to test those paths for other install media 14:28:27 * mclasen will be in an airplane during the fesco meeting 14:28:33 netinstall ISO, server ISO, etc. 14:28:36 Ah no, it's not a good point, because now we are promoting path (d) created neither initial user nor root password in anaconda. It does increase the QA matrix. ;) 14:28:45 I don't htink it's fair to have a full discussion here - let's just agree to take it back to fesco, and try to show up this time :-) 14:28:47 IMHO (1) we should reapproach FESCo and explain where we see some miscommunication/misunderstanding; and (2) we should commit to having one or two knowledgeable WG folks at the next FESco if possible (or hold off until we can) 14:28:59 otaylor: *jinx 14:29:00 Where previously they could usually trust that such a test passing in any media was a valid test for all of them 14:29:07 stickster: +1 14:29:12 +1 14:29:15 stickster: +1 14:29:29 even if we need to have some more tests... it's not impossible for folks here to help QA with that part 14:29:39 ack 14:29:46 it's a finite (and relatively small) # of tests to deal with 14:29:52 and most of this stuff is automated AIUI 14:30:02 everyone loves code ;-) 14:30:13 What's our plan to get this onto FESCo's agenda? Resubmit the change proposal without implementing its requests? Or create an issue tracker issue...? 14:30:20 I *would* like to hear a plan around the netinstall question at the FESCo meeting, thoguh 14:30:26 mcatanzaro: reopen https://pagure.io/fesco/issue/1803 14:30:33 mcatanzaro: I can just add it to the meeting agenda upon your request. 14:30:53 Please reopen 1803 and I'll add it 14:31:08 who can be at the Friday session? 14:31:23 I should be able to be there 14:31:34 (Even if the netinstall plan is "we will discourage use of netinstall", I'd like to know the plan) 14:31:41 I should be as well 14:31:53 * otaylor can be there too 14:31:56 I can try to be there, but can't promise it. 14:32:04 OK, it shouldn't take more than 1-2 people to advise 14:32:40 the netinstall question sgallagh asks is important too -- we don't need to settle that here but let's use #fedora-workstation to discuss/figure out 14:33:13 I was going to implement by modifying the workstation kickstart, which would mean absolutely no change to netinstall 14:33:25 kalev: otaylor: So I can assume you guys are +1 with the plan ;-) 14:33:28 for large scale deployments I thought not having user creation in the installer was seen as a big plus 14:33:54 Basically the two options are (a) make netinstall work like the live installer, or (b) make it work like all the other editions' installers (no change). adamw had some arguments in favor of (a), but (b) is easier to implement and seems maybe less-controversial. 14:33:55 because you can create an image and have the user create their accounts afterwards 14:34:31 cschalle: Well, for large-scale deployments, the answer is usually "use a domain controller", but sure. 14:34:43 yeah, ansible, fleet commander, or what have you 14:34:45 cschalle: if large deployment is oem laptops :-) 14:34:56 kickstart... 14:35:49 For large deployments, there will be a kickstart, and people are *not* typing in 1000 user passwords or root passwords manually - so really the only question is someone who decides to download the net install instead of the workstation because someone told him it was a good idea 14:36:24 the internet will always be there to hand out loaded six-shooters to those with willing feet 14:36:35 I think I'd go with (a) and try to make netinstall as similar to the live installer as we can have it 14:37:00 kalev: that would require anaconda engineering to at least a small degree, right? 14:37:12 I don't know 14:37:15 To be clear, there are two netinstalls to consider: a Workstation-branded one and the "generic" one. 14:37:39 stickster: how does the agenda look - should we continue this topic, or discuss on #fedora-workstation? 14:37:39 I'm perfectly happy with you deciding "no changes at all to the generic one". 14:37:41 sgallagh: do we have such a thing? 14:37:45 I think it's fair to say that this change should NOT make the generic netinstall behave like workstation. we need the generic one for other desktops/server 14:37:54 otaylor: Thanks. I would like to get to our other item 14:38:02 stickster: We do 14:38:06 TIL 14:38:12 as for the workstation netinstall, I'd try to make it as close as possible to workstation netinstall, if it's possible 14:38:17 sgallagh: Can you reopen https://pagure.io/fesco/issue/1803 please? (Thanks for attending today!) 14:38:23 ack 14:38:30 if 'large deployment' was a use case we cover, I would expect it to be mentioned on getfedora.org - no sign of it there 14:38:58 #agreed reopen FESCo issue #1803, and at least cschalle/kalev will attend FESCo on 2018-01-19 to advise and get a new, fully informed decision 14:39:15 #action sgallagh reopen https://pagure.io/fesco/issue/1803 14:39:24 #action cschalle kalev attend FESCo on 2018-01-19 14:39:36 Reopened 14:39:51 #topic Noto color emoji 14:39:54 thanks sgallagh ! 14:39:58 #link https://pagure.io/fedora-workstation/issue/31 14:41:14 I think the idea is that the Noto Color Emoji font is better technically so the proposal or question is should we not use it as the default for F28 14:41:32 Noto looks much better IMO 14:41:37 We've previously decided not to use it, though 14:41:42 why was that? 14:41:46 Though we had some discussion about the color fonts before 14:42:04 I think because Android uses it? hadess really wants to use Emoji One. 14:42:14 mcatanzaro: well at least for F27 - I forget exactly 14:42:26 mcatanzaro: right 14:43:05 pretty sure fedora-workstation and the exact same query for f27 have the details of "why" 14:43:15 It was suggested to open a ticket later with details about advantages of the Noto font if we wanted to consider that 14:43:35 juhp: "later" being "in at least 2 cycles" 14:43:49 was it? 14:44:00 I don't remember that 14:44:05 I don't remember the wait 2 cycles either 14:45:09 https://bugzilla.redhat.com/show_bug.cgi?id=1496761#c17 <- "within the next year" 14:46:01 i'm sure there's also discussions about this on the fedora-desktop list, and possibly other places 14:46:44 Well, one more cycle would be around this summer... are you confident that Emoji Two will be usable by then, and that it will solve the problems brought up in the ticket? Looks like Noto is clearly the better option as of today. 14:47:09 hadess: the bug points to https://lists.fedoraproject.org/archives/list/desktop@lists.fedoraproject.org/thread/3IS6VEB53SJSQCY5W223LR2K7ZJ4AH3T/#3IS6VEB53SJSQCY5W223LR2K7ZJ4AH3T which was not what I'd call a discussion 14:47:13 emoji two doesn't exist in a format we can use 14:47:51 "emoji two doesn't exist in a format we can use" <-- exactly why I'm skeptical about continuing with Emoji One 14:47:53 so mfabian didn't test emoji two i don't think 14:48:27 I don't believe emojitwo is any better technically correctly 14:48:31 at least last I heard 14:48:39 so far all the arguments seem to be "it was decided to use the other font, please don't ask again" 14:48:42 s/correctly/currently/ sorry 14:48:43 one of the main problems with emoji one (broken rendering of some sequences), I have addressed in gtk 14:48:43 ... which I don't really understand :) 14:49:07 And, Emoji One is no longer free upstream, we are stuck on an old version. 14:49:19 right 14:49:32 hadess: is the problem here that we're trying to change the font in fedora, whereas you'd like to select the default font within gnome? 14:49:35 But if there are going to be improvements here within the next year, I understand switching to Noto now and then back later would not be great. 14:50:07 kalev: the default isn't set in gnome, but in fontconfig 14:50:16 kalev: i'd like the default font to be the same in both 14:50:27 ok, fair enough. isn't juhp fontconfig maintainer and can change it? 14:50:40 kalev: we chose emoji one for branding purposes, so a gnome desktop doesn't look like android 14:50:41 not me, tagoh 14:51:03 hadess: I dont' think the fontconfig upstream is the place for choosing default fonts 14:51:31 kalev: akira is the fontconfig maintainer, but the emoji support there was by and large done be behdad 14:51:34 we can override the priorities in our font packages 14:51:36 * stickster notes: *** 5 minutes left in meeting, since chair has to report elsewhere at that point *** 14:52:21 otaylor: it's what it is, there's no way to override it somewhere else, and given that it can be overridden just the same in fedora, i don't see the point of discussing this particular point 14:52:59 we recognise that noto is a better font technically, but asked for time to see if contributors would appear now that emoji one/two are defaults for gnome (and more generally linux) desktops 14:53:13 I hate to say it, but I don't find "so a gnome desktop doesn't look like android" very compelling 14:53:22 rdieter++ 14:53:22 mcatanzaro: Karma for rdieter changed to 4 (for the f27 release cycle): https://badges.fedoraproject.org/tags/cookie/any 14:53:30 hadess: Maybe I misunderstood your comment - what I'm saying is that the decision about what the font should be should be done at a level that can make decisions about overall "desktop appearance - either GNOME or Fedora 14:53:49 https://cgit.freedesktop.org/fontconfig/commit/?id=c41c9220181b203d1cf1f6435f6e3735cb7c84ac this change made it so that fontconfig prefers noto actually, no ? 14:53:56 otaylor: we asked fontconfig maintainers to do that 14:54:14 mclasen: yes, if it is installed manually 14:54:19 otaylor: We currently do use upstream fontconfig defaults though (Deja Vu) 14:55:00 We'll we're out of time... shall we continue the discussion at the next meeting? 14:55:07 rdieter: if "so a gnome desktop doesn't look like android", then please tell matt about this so we can cut the branding discussions short 14:55:11 rdieter: mcatanzaro: I kind of agree when it comes to a detail like this (although appreciate the larger view across the desktop) 14:55:28 ok, so what would be our deadline to make this change for F28? meaning how long can we delay this decision to give emojitwo a chance to prove itself? 14:55:29 hadess: I think you're being overly dramatic 14:55:41 Or vote in the ticket... looks like likely options are (a) switch to Noto now, or (b) wait until summer and switch then if Emoji Two has not improved by then. 14:55:51 cschalle: I think we should allow a bit more time, i.e. until F29 Rawhide branch 14:55:55 imho, technical superiority should trump (simple) branding 14:55:58 we could ask for mattdm to pitch in here if needed 14:55:58 I think ideally it should be Change 14:56:15 rdieter: i don't think it's being dramatic, no, it's completely within the realm of branding 14:56:22 rdieter: there *is* a wrinkle to that, where we would constantly be ceding ground to projects run by large companies with deep pockets 14:56:33 I was naively hoping for f28 14:56:33 OTOH that's an ironic thing for a Red Hatter to say ;-D 14:56:40 hadess: sure it's branding, but the "cut the branding discussions short" part 14:57:18 I suggest we need to continue the discussion next meeting here 14:57:20 there's way more to the branding discussion than this very small part 14:57:33 and/or on list as a *real* discussion 14:57:37 do we have a UI branding strategy for fedora? 14:57:50 rdieter: this is a much more effective piece of branding than slapping logos around, and i can see it being sold short 14:58:03 I have to close this meeting, apologies folks -- let's take it to #fedora-workstation 14:58:18 Bye! 14:58:23 thanks for the discussion 14:58:28 #info No decision point reached, this issue needs more discussion and consensus from the WG and contributors 14:58:34 hadess, but it is a 'branding' issue with consequences beyond branding 14:58:36 stickster: wait, this meeting is only pretend discussion ? 14:58:50 mclasen: vs. the thread I linked earlier, not vs. this meeting ;-) 14:58:51 * mclasen wasted his time 14:59:09 mclasen: We clearly don't have consensus yet, let's retake the discussion at the next meeting 14:59:11 Thanks for coming, everyone 14:59:17 I assume having a broader discussion that includes more than just present WG members could be helpful 14:59:20 thanks for chairing stickster 14:59:25 #agreed Resume discussion on emoji issue next meeting 14:59:30 #endmeeting