18:13:36 #startmeeting Go SIG meeting 18:13:36 Meeting started Mon Dec 21 18:13:36 2020 UTC. 18:13:36 This meeting is logged and archived in a public location. 18:13:36 The chair is alexsaezm. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:13:36 Useful Commands: #action #agreed #halp #info #idea #link #topic. 18:13:36 The meeting name has been set to 'go_sig_meeting' 18:13:42 it worked! 18:14:35 .hello2 18:14:36 eclipseo: eclipseo 'Robert-André Mauchin' 18:15:11 use #chair username for active participants 18:15:39 nim jcajka QuLogic Eighth_Doctor meeting time 18:15:42 #chair alexsaezm 18:15:42 Current chairs: alexsaezm 18:15:50 #chair eclipseo 18:15:50 Current chairs: alexsaezm eclipseo 18:16:07 hey 18:16:52 hi 18:17:27 AFAIK we don't have relevant stuff to talk about (correct me if I'm wrong please) so we can move to the open floor 18:17:57 and... not really sure if I should be leading this meeting lol :D 18:18:17 wait for jcajka I have question for him 18:18:25 sure :) 18:19:28 It looks like we're going to have a get a move-on to support Go modules 18:19:32 anyone have any idea how we're going to do that? 18:20:08 Right, that's a relevant thing to talk about 18:20:37 Do we know how it will affect us? 18:21:04 currently we remove vendoring 18:21:24 It depends on nim tools and if they are ready 18:21:46 I don't think we can target F34 for that 18:22:11 nim tools? sounds rad :D 18:22:21 We would need to rebuild the 1600 packages or so 18:22:23 the 100% modules migration is on 1.17, isn't it? 18:23:24 Go 1.17 (August 2021) will remove the GO111MODULE setting and GOPATH mode entirely, using module mode always 18:23:29 ^ https://github.com/golang/go/wiki/GOPATH 18:23:30 alexsaezm: https://pagure.io/modist 18:23:53 thanks eclipseo 18:24:17 alexsaezm: well they already use modulet but in 1.17 they plan to drop gopath entirely 18:25:17 I'm not super enthousiast on having to rebuild the entire stack from scratch like I did last time 18:27:47 you know, this is a remarkably insensitive thing for the Google Go compiler team to do 18:28:06 such a massive break in behavior blows up basically everything 18:28:14 and it's also going to be inconsistent with gcc-go 18:28:18 well it's not like they listen to downstream 18:28:24 which, as far as I can tell, hasn't been primed for this change either 18:29:05 well we don't really support gcc-goc neither they do I believe 18:33:10 the Google Go compiler team maintains both ports 18:33:15 golang and gcc-go 18:34:38 alexsaezm: there are a couple of examples of Go modules here https://copr.fedorainfracloud.org/coprs/nim/go-modules/builds/ 18:35:44 thanks a lot, I was trying to figure out what exactly modist works and does and how this situation really affect us to... build an example and understand it, but if nim already did that, then awesome :D 18:36:20 the syntax is a bit complicated and requires more work than with GOPATH because we have to provide che go.mod if it is missing or if the package have to be split due to cyclic deps 18:36:54 there's also the fiddling we'll have to do with go.sum 18:39:35 also i believe RHEL9 will be based on F34, so it will lag behind if we manage to introduce it in F35 18:40:06 RHEL 9 will be openly developed as CentOS Stream 9, so hopefully we can get those changes to land before the cutoff for GA stabilization 18:40:17 but ideally, we should get this in for F34 if we can 18:41:16 isn't it a little late? I mean, is a mass rebuild and I think the proposals are close or be closed soon 18:42:15 system-wide changes deadline is the 29th 18:42:43 Eighth_Doctor: I can't rebuild the 1600 packages before F34 release 18:44:06 also now to drop some test we have to patch them manually which will cause a massive increase in tie spent by package 18:48:55 Just to have a good picture of the situation... if I go to, for example, podman (first package came to mind) and set GO111MODULE=on... that would be the change, wouldn't it? 18:49:17 I mean in the specfile 18:49:22 basically, yes 18:49:41 Ok, let's break things (on local) 18:49:46 there's a macro to override to remove the GO111MODULE=off behavior 18:50:05 I added it a while ago when I was trying to package up Hashicorp Vault 18:50:36 but if you are unbundled that requires repackaging the entire deps 18:50:49 anduse rew macros to handle it 18:50:58 yeah, I hadn't done that for my attempt 18:51:42 if you are vendored it's not a n issue, I already used GO111MODULES in EPEL8 18:52:31 ah 18:52:39 Yeah, RHEL is vendored right now for example, that why I'm having a hard time to grasp this :( 18:55:13 ok, podman broken. got it 18:55:23 I've got to go :/ Lets discuss this with Nim and jcajka next meeting if they are available 18:55:41 sounds good 18:55:51 got it 18:55:51 we should probably just make this a persistent agenda item until we solve this 18:55:51 I haven't touch podman ever, I don't know whi maintains it 18:56:05 *who 18:56:15 Eighth_Doctor, is there a place where I should write that down? 18:56:38 https://pagure.io/GoSIG/go-sig/issues 18:57:14 Really gotto go now 18:57:19 bye eclipseo 18:57:24 bye eclipseo 18:57:25 Don't forqet to #endmeeting 18:57:27 bye 18:57:39 well, since there's nothing else to talk about... 18:57:54 do we want to end the meeting? 18:58:16 I think so :) 18:59:11 I'll do it :) 18:59:24 #endmeeting