13:00:45 #startmeeting silverblue 13:00:45 Meeting started Mon Sep 3 13:00:45 2018 UTC. 13:00:45 This meeting is logged and archived in a public location. 13:00:45 The chair is mclasen. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:00:45 Useful Commands: #action #agreed #halp #info #idea #link #topic. 13:00:45 The meeting name has been set to 'silverblue' 13:00:52 #topic roll call 13:00:57 morning 13:01:05 .hello2 13:01:06 sanja: sanja 'Sanja Bonic' 13:01:09 hey kalev! 13:01:13 .hello sinnykumari 13:01:14 ksinny: sinnykumari 'Sinny Kumari' 13:01:19 Hi everyone! 13:01:32 might be light attendance today. Its the rare us-only vacation day... 13:03:02 o/ 13:03:10 .hello2 13:03:11 misc: misc 'None' 13:04:04 .hello2 13:04:08 rishi: rishi 'Debarshi Ray' 13:04:19 .hello2 petersen 13:04:20 juhp: Sorry, but you don't exist 13:04:34 :) 13:04:40 ok, some folks are here :) 13:04:50 lets start by collecting agenda topics 13:04:53 another shock ;) 13:05:03 :) 13:05:14 I have one item in pagure thats marked as meeting 13:05:32 .hello2 13:05:33 juhp: Sorry, but you don't exist 13:05:47 since we have kalev back, we could discuss the state of gnome-software vs rpm-ostree for f29 13:06:10 and since debarshi is here, we could discuss the state of toolbox containers for f29 13:06:14 anything else ? 13:06:19 great 13:06:54 I wanted to discuss F29 Silverblue Test Day, we have a ticket opened for it 13:07:00 not for me for now 13:07:20 oh, right. test day 13:07:28 * mclasen adds that to the agenda 13:07:36 lets get started then 13:07:59 #topic yelp in silverblue, https://pagure.io/teamsilverblue/issue/31 13:08:26 somebody pointed out that we dropped yelp, with a bunch of other applications 13:08:54 oh, how did that happen? did it get dropped from regular workstation or just from silverblue? 13:08:54 which is not a problem for flatpaked apps from flathub, since yelp is in the runtime, so their help works 13:09:20 there was a bunch of apps dropped (there's a discussion about it on discourse) 13:09:24 let me find that 13:09:55 https://discussion.fedoraproject.org/t/changes-in-fedora-silverblue-29/325/28 13:10:07 well, drop the /28 at the end 13:10:50 * kalev nods. 13:11:17 I think we can get all of those apps built as flatpaks in koji, since owen is going to unveil his flatpak build support this week 13:11:45 the reason that yelp was called out in the ticket is that we need it for the initial-setup welcome part 13:11:57 after the first login 13:12:19 oh wow, flatpak build support this week? awesome stuff 13:12:20 and since we won't have preinstalled flatpaks in f29, building it as a flatpak is not going to fix this case 13:12:43 #info building flatpaks in koji is expected to start this week 13:13:02 so I would suggest that we should put yelp back in the base image for f29 13:13:27 * kalev agrees. 13:13:33 +1 13:13:48 +1 13:13:53 +1 (can I vote?) 13:14:02 +1 13:14:20 +1 13:14:38 alright, I think thats a clear vote 13:14:53 now, how do I record that for the minutes 13:14:59 * mclasen looks for the cheat sheet 13:15:18 +1 13:15:29 #agreed We will put yelp back in the base image for F29 13:16:15 #action mclasen will write the patch to put it back 13:16:40 #topic test day for silverblue 13:17:04 ksinny: what timeframe are we looking at ? 13:17:09 #link https://pagure.io/teamsilverblue/issue/28 13:17:40 mclasen: I think it will nice to have it soon after F29 Beta release 13:18:13 But, it's totatlly upon when Silverblue community memebers will be free 13:18:24 so... September 20 or so ? thats a thursday 13:19:00 that would be my suggestion 13:19:16 +1 from me, if it works for QA 13:19:30 20th September works for me too. 13:19:42 I am not very particular. 13:19:56 Works for me as well 13:19:57 right, we should perhaps confirm with qa 13:20:24 Are those 4 test cases enough or they can be increased now? 13:20:32 yeah, if we are fine with 20th Septemeber then I will drop a comment to QA ticket 13:20:47 paragan: where are the test cases ? 13:20:54 http://testdays.fedorainfracloud.org/events/39 13:21:01 Atomic Workstation 13:21:03 I'm sure we can add a few more between now and September 20 13:21:15 Yeah, that will be great 13:21:45 mclasen: can we have all new tests we want to get tested listed in our Test day link? 13:22:02 ksinny: will you announce the test day when the date is finalized ? 13:22:06 Later, QA can form proper Test Case with that 13:22:28 #link http://testdays.fedorainfracloud.org/events/39 13:22:57 whats the best place to collect test case input, the test day ticket ? 13:23:37 mclasen: Sure, I can work on writing a Fedora Magazine article for Silverblue Test Day. Will first ask in silverblue-apac meeting to see if anyone else wanted to pick it 13:23:53 excellent, thanks! 13:24:11 #action ksinny will publicize the silverblue test day for (tentatively) September 20 13:24:39 #topic status of rpm-ostree support in gnome-software for Fedora 29 13:24:48 kalev: putting you on the spot on your first day back :) 13:25:28 eeks :) 13:26:42 soooo, I have a WIP item that is adding package layering support to gnome-software 13:27:32 before I went away, I was playing with some rpm-ostreed patches to add the needed dbus API to make it possible 13:27:38 how close are you to having that working ? 13:28:31 I think it should be fairly close, pretty sure this should be working before F29 enters the freezes at least 13:28:42 ah, nice 13:28:59 the main place where this will show up in the ui is add-ons, I think ? 13:29:21 do we have the metadata we need to populate the ui for that ? 13:29:56 the most visible thing that should change is that the front page gets back to fully functional, with categories showing up and all that 13:30:23 that's nonfunctional out of box right now because we don't show up anything that can be layered 13:30:24 oh, is that related to package layering ? 13:30:26 yeah 13:30:51 it's all empty there because there are no apps to show up 13:30:53 do we want to have package layering on the front page ? 13:31:04 that would be rpm-packaged apps showing up again? 13:31:09 * kalev nods. 13:31:42 could send the wrong message, and make everybody create big layers ? 13:32:00 I think we need to try to do something for the next cycle where flatpak souces get preferred treatment on the front page 13:32:04 Allan had some ideas for that 13:32:29 but at this stage, I think anything is better than empty front page :) 13:32:29 Can we make an informational layover screen that explains and allows to tick a checkbox to enable flatpak and show the apps to layer but yu gotta click it away first and enable it? 13:32:45 right. better treatment of multiple sources for the same app is on my wishlist for next cylce 13:33:16 sanja: I think this is dependant on getting flatpaks built in Fedora 13:33:32 once we have that, we can add them either by default or opt-in or whatever 13:33:37 hopefully, this week 13:34:04 also, layering requires a reboot to have a package present, or use the livefs feature, so the UX will ask to restart I guess ? 13:34:05 ok 13:34:09 I am not sure if owen was comfortable with enabling the registry as source by default 13:34:14 why not? 13:34:20 but it is a good thing to add to the test day testcases 13:34:45 just because its all new, and using less tested code paths in flatpak (oci...) 13:34:48 also, we need to have packages as flatpak once they are possible to create, not sure if we have a plan for that 13:34:56 but if we find it works, we should enable it 13:36:04 kalev: anything else to add to rpm-ostree status ? what about rebases ? 13:36:14 * sanja has to go to school pickup - if there's any action items for me, let me know afterwards. Bye for now! 13:36:21 bye, sanja 13:37:54 mclasen: I hooked up distro upgrade support in gnome-software before I left, so that it uses the same metadata as regular distro upgrades 13:38:21 ok. I forget - is this something we can test before f29 is released, or do we lack metadata for that ? 13:38:23 and then tries to rebase to a hardcoded ostree branch name, just replacing the distro number in there 13:38:53 this needs latest gnome-software for F28. I can try setting up a COPR for that. 13:39:06 it would be nice if we could figure out a way to make it testable for the test day 13:39:12 * kalev nods. 13:39:29 can you look at that ? 13:39:32 sure, will do 13:39:52 #action kalev will look at making rebase in gnome-software testable for the test day 13:40:31 #info kalev hopes to have package layering working in gnome-software in time for the final freeze 13:40:53 #topic toolbox containers 13:40:58 * rishi waves 13:41:11 rishi: want to give a quick update ? 13:41:40 So, I put up a Git repo for the toolbox script and Dockerfile: https://github.com/debarshiray/fedora-toolbox 13:42:32 #info toolbox container materials here: https://github.com/debarshiray/fedora-toolbox 13:42:35 We are still ironing out the bugs in rootless podman. eg., it currently fails to map the UID from the host to the container. 13:43:02 But you can create the container, enter it and get a shell. 13:43:34 have you gotten some people to try it out ? 13:43:41 I did 13:43:53 Hopefully, we will have some kind of fedora-toolbox image in the Fedora registry as the base for creating these containers. 13:44:01 That's tracked here: https://pagure.io/releng/issue/7698 13:44:37 Although, I have a feeling that the releng ticket has gotten derailed a bit. 13:44:55 mclasen: misc already sent a pull request! 13:45:05 misc: +1 13:45:18 But I think it can only realistically be advertised once the UID mapping is fixed. 13:45:38 eg., running rootless, one can't even 'sudo dnf ...' inside the toolbox, which is a bummer. 13:46:06 but the sudo approach works for now ? 13:46:40 There's also a thread about reviewing the Dockerfile for creating the base toolbox image: https://discussion.fedoraproject.org/t/an-image-for-a-fedora-toolbox-container/286/5 13:46:44 Comments welcome, etc.. 13:47:05 mclasen: Yes, the sudo approach works for now, but that's not so easy to set up. :/ 13:47:50 Good thing is that gscrivano is already chasing down the runc bug that prevents us from moving forward with the UID mapping: https://github.com/opencontainers/runc/pull/1862#issuecomment-417784925 13:48:05 that is great news 13:48:59 so do you envision we will get this on people's systems ? 13:49:07 In short, we found that it works with 'podman run', so that validates our approach, but due to various details it breaks with 'podman create' and 'podman exec', which is what the toolbox uses. 13:49:13 I guess the ultimate goal would be to have this preinstalled somehow ? 13:49:24 mclasen: Yes, I am hopeful. 13:49:41 I guess it will be part of the base image? Like podman, buildah, etc.. 13:49:54 is there going to be a package that we need on the host ? 13:50:19 Yes, we do. RPMs are our currency, so ... :) 13:50:30 Although, I am not sure about the naming. 13:50:48 whats the current candidate ? 13:51:10 It's fedora-toolbox.git at the moment. But I am not attached to it or anything. 13:51:14 When we discussed toolbox containers with the Fedora CoreOS gang, there was a suggestion ... 13:51:59 ... to fork off coreos/toolbox.git and work on that, but coreos/toolbox.git has a bunch stuff around governance, etc. which I wasn't sure what to do about. 13:52:13 So I threw in fedora-toolbox to avoid too much bikeshedding. 13:52:30 seems fine to me, fwiw 13:52:56 Although, fedora-toolbox does seem to assume that both Silverblue and Fedora CoreOS will use the same thing, because it takes over the "fedora" namespace. 13:53:40 Is it intended for both or ? 13:53:47 For those reading along, here's some context from the Fedora CoreOS side of the issue: https://github.com/coreos/fedora-coreos-tracker/issues/38 13:53:58 juhp: Hopefully/ideally, yes. 13:54:00 we discussed that a bit in the coreos meeting last week 13:54:04 okay 13:54:16 But CoreOS' requirements are a bit different from ours. 13:54:17 seems we have somewhat different use cases, but quite a bit of overlap too 13:54:25 Yeah. 13:54:38 * mclasen needs to run to the next meeting... some vacation day :( 13:54:59 My understanding is that we will use the same underlying bits - podman, buildah, etc.. 13:55:22 But, for example, CoreOS is fine with asking the operator to type in the root password to get the container ... 13:55:31 I'll end this meeting, and we can all move to #silverblue to continue chatting ? 13:55:43 ... which is not ideal if we want people to get a toolbox prompt in Silverblue easily. 13:55:45 thanks for the update, rishi! 13:55:52 mclasen: Yes, sure! 13:55:55 #endmeeting