19:00:06 #startmeeting Go SIG meeting 19:00:06 Meeting started Mon Jan 30 19:00:06 2023 UTC. 19:00:06 This meeting is logged and archived in a public location. 19:00:06 The chair is alexsaezm. Information about MeetBot at https://fedoraproject.org/wiki/Zodbot#Meeting_Functions. 19:00:06 Useful Commands: #action #agreed #halp #info #idea #link #topic. 19:00:06 The meeting name has been set to 'go_sig_meeting' 19:00:17 #topic Roll Call 19:00:24 Hi there! :) 19:00:42 .hello gotmax23 19:00:42 gotmax[m]: gotmax23 'Maxwell G' 19:01:37 .hello copperi 19:01:41 copperi[m]: copperi 'Jan Kuparinen' 19:02:19 mikelo eclipseo_ Fale: Are any of you around? 19:03:49 * alexsaezm will give until 05 so people can arrive :) 19:03:57 Ack 19:05:37 we have a bunch of stuff in the Go SIG issue tracker. Any preference? 19:05:50 My Change Proposal :) ? 19:07:01 Sounds like an awesome starting point :D 19:07:05 (https://fedoraproject.org/wiki/Changes/Mass_Retire_Golang_Leaves is the link) 19:07:22 #topic Leaf Go library packages 19:08:10 Also, do mind doing #chair gotmax[m] ? 19:08:10 It look good to me 19:08:36 #link https://fedoraproject.org/wiki/Changes/Mass_Retire_Golang_Leaves 19:08:44 #chair gotmax[m] 19:08:44 Current chairs: alexsaezm gotmax[m] 19:08:47 did it work? 19:08:49 it did! 19:08:50 yay 19:08:54 👍️ 19:08:57 * alexsaezm is always afraid of zodbot 19:09:06 Don't be 19:09:33 * gotmax[m] hopes that zodbot won't rise up and kill us all 19:09:50 haha 19:10:01 gotmax (He/Him): I checked the proposal and I really liked it 19:10:10 thanks :) 19:10:42 I wanted to bring it up in the meeting before I submitted it 19:11:35 does anyone have any specific questions or concerns? 19:11:36 LGTM, can't think of anything bad to say :D 19:12:42 cool. I'd also like to highlight the opt-out process. 19:13:07 It's at the end of the https://pagure.io/GoSIG/go-leaves/tree/main README 19:13:42 I really like that 19:13:57 that's a clever idea 19:14:17 I wanted something that was easy to parse and was auditable 19:15:16 I guess it's harder for others than creating a wiki page and allowing them to put package names there, but this makes my job easier :) 19:15:44 Anyways, I'll submit this proposal to releng and then submit it to the Program Manager 19:16:52 thank you very much 19:16:55 gotmax (He/Him): ++ 19:18:06 anything else with this topic? 19:18:44 I don't think so 19:19:34 Re. other topics, I think we've covered merging specfiles, golang-race was already removed, and I don't have any updates on %ix86 (I'm working on this first). 19:19:46 I was gonna say exactly that 19:19:51 :) 19:19:58 docker brokenness then? 19:20:17 I don't have much here either 19:20:28 sergiomb: has been maintaining moby-engine 19:20:36 Meh, I guess I have some 19:20:49 #topic Discuss current docker/moby/containerd ecosystem brokenness 19:20:52 #link https://pagure.io/GoSIG/go-sig/issue/43 19:21:35 A non-packager has been submitting PRs to update containerd that I've been merging (Robert is the primary maintainer). 19:21:39 Maybe I'll offer to sponsor them 19:22:25 that is great :D 19:22:28 +1 19:23:06 golang-github-docker and golang-github-docker-cli have been updated to the next major version (v20 -> v22). They are a couple betas behind. I think the dependencies are in a little bit better shape, but still many are out of date and/or FTBFS. 19:24:01 Once v22 is officially released, we should update moby-engine. I can try to help out a little with that. 19:24:25 There's been a change to how it handles bundled dependencies. We might also be able to unbundle it. 19:24:26 * bundled dependencies in v22. We 19:24:48 so should they be vendored ? 19:25:55 moby-engine is currently built from vendored deps. golang-github-docker and golang-gthub-docker-cli contains the same code but not the binaries. They're used as libraries for other go packages. 19:26:25 copperi[m]: not necessarily 19:26:40 unbundling will require somewhat large packaging changes 19:28:02 When it comes to unbundling, I'd like to suggest that you think whether unbundling adds any value and whether that is worth it 19:28:34 Well, it's more in the spirit of Fedora packaging guidelines. Those large vendor tarballs are a bit dubious. 19:29:05 packaging guidelines can be changed 19:29:20 I know it's very much against the spirit of the Fedora packaging guidelines 19:29:39 but the packaging story for go, rust, node, etc. is not getting any easier 19:30:02 we are already maintaining the unbunled golang-github-docker anyways, because other packages need to import its code. 19:30:25 then unbundling that could make sense 19:30:56 They've gotten rid of unbundled node libraries, and I don't think that's really increased nodejs packaging 19:33:12 * gotmax[m] goes to check something. brb 19:34:40 also, moby-engine bundled tini-static, which I really don't like. it's used for docker run --init. it should be possible to add Requires: tini-static package and add --init-path /usr/bin/tini-static to dockerd.service 19:39:16 is there anything else to talk about this topic? 19:39:27 * gotmax[m] doesn't think so 19:39:43 then... 19:39:44 #topic Open floor 19:41:08 I just have one question. I think eclipseo_ it's quite busy these days I have a package review that I would love to merge. Is anyone interested it taking a look at https://bugzilla.redhat.com/show_bug.cgi?id=2102388 ? 19:41:55 he was reviewing it 19:43:25 I can maybe take a look later this week. Anyone else is free to go ahead. 19:43:40 thank you 19:45:47 I've filed https://pagure.io/releng/issue/11253 to get releng approval 19:47:48 hi, yes I'm trying to maintain docker . the problem of unbundle things like go packages is that we will have a bunch of packages, yet another day I approved one package that the code was 50 lines or less , so is not about the spirit 19:49:22 Just in case, this might be relevant for you https://fedoraproject.org/wiki/Bundled_Libraries 19:51:11 Hmm, I didn't know about that page, but looks informative 19:52:32 It's a hot topic and I have mixed feelings but I think it's a really interesting take on the matter 19:54:43 I think that having dependency trees this large is a problem that's orthogonal to bundling. It makes it hard to audit and way too easy for some hidden transitive dependency of a transitive dependency of transitive dependency to push a malicious update and rm -rf your whole system. 19:55:46 yeah, I think we are reaching if not already reached the point of "too many" 19:56:28 RHEL does the opposite but the surface is way smaller 19:59:08 Is there anything else we want to discuss? 19:59:19 * gotmax[m] shakes his head 20:00:12 in that case... I think we can call it a day :) 20:00:25 thank you a lot and see you in the next one! 20:00:25 #endmeeting