20:31:35 <prometheanfire> #startmeeting requirements 20:31:36 <openstack> Meeting started Wed May 22 20:31:35 2019 UTC and is due to finish in 60 minutes. The chair is prometheanfire. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:31:37 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:31:39 <openstack> The meeting name has been set to 'requirements' 20:31:54 <tonyb[m]> O/ 20:31:56 <prometheanfire> #topic rollcall 20:32:03 <prometheanfire> tonyb, prometheanfire, number80, dirk, coolsvap, toabctl, smcginnis, dhellmann ping 20:32:06 <prometheanfire> o/ 20:32:35 <tonyb[m]> smcginnis: is in Barcelona ATM so probably won't be here 20:32:48 <tonyb[m]> ... as is dhellmann 20:32:51 <prometheanfire> yep 20:34:17 <prometheanfire> #topic controversies in the queue 20:35:22 <prometheanfire> I think we are generally fine here 20:36:07 <tonyb> Yeah 20:36:32 <prometheanfire> #topic email the list for issues we had a couple of weeks ago 20:36:36 <prometheanfire> #link https://etherpad.openstack.org/p/reqs-py2-caps 20:38:15 <tonyb> I'll make time to add my thoughtts today. 20:38:21 <prometheanfire> ok, thanks 20:38:25 <prometheanfire> #topic open floor 20:39:08 * tonyb has 2 things 20:39:35 <tonyb> 1) ... following on from the courtesy pings discussion on the m/l 20:39:49 <tonyb> I think we should clean out our list 20:40:03 <prometheanfire> our list? 20:40:18 <tonyb> I suggest we update the wiki with "ping list started Jun15" and opt in again 20:40:33 <tonyb> then from Jun 15 we stop pinging the existing list 20:40:39 <prometheanfire> oh, for roll call, ya 20:40:54 <tonyb> as I don't see number80, tobctl or coolsvap at all 20:41:08 <prometheanfire> :D 20:41:54 <tonyb> Shall I do a picard ... and make it so? 20:42:09 <prometheanfire> please do 20:42:15 <prometheanfire> the other item? 20:43:03 <tonyb> sorry multitasking 20:43:09 <prometheanfire> np 20:43:34 <tonyb> evrardjp: brought up that our 'check-uc' jobs install things but don't build wheels 20:43:43 <smcginnis> tonyb[m]: Actually not in Barcelona. Plans were changed for me. 20:43:49 <tonyb> (which I need to investigate but assuming that's right) 20:43:57 <tonyb> smcginnis: Oh dear! 20:44:12 <smcginnis> Yeah, it's been an interesting few days. 20:44:33 <tonyb> so do we feel about adding a LOCI job or similar to build the wheels in our check and gate pipelines? 20:44:39 <dirk> o/ 20:45:02 <prometheanfire> tonyb: is it our job to build/ship wheels? 20:45:19 <prometheanfire> I understand we may want to check if they are available 20:45:29 <tonyb> prometheanfire: no but it is our job to ensure things work and are installable 20:45:56 <prometheanfire> right, adding a job for checking that makes sense 20:46:00 <tonyb> IMO adding a job like this will have our consumers pain so it's something we shoudl consider 20:46:21 <dirk> so the issue was that osh got a constraints update where the wheels was not buildable? 20:46:25 <dirk> on some distribution? 20:46:32 <dirk> or on any distribution? I guess I'm missing some detail 20:46:52 <tonyb> dirk: Yeah, it looks like perhaps only centos was/is broken 20:47:47 <tonyb> ... evrardjp says: M2Crypto fails to build its wheels under centos/ubuntu in LOCI 20:48:09 <prometheanfire> tonyb: so would we be building wheels (I assume so) and if so would we be shipping them (that's releases / infra's job right?) 20:48:11 <tonyb> so it's worth looking at why (and he provides a log) and if it's LOCI specific 20:48:26 <prometheanfire> if we ship them how do we not become the mirror everyone uses 20:48:31 <tonyb> prometheanfire: we'd just build them and discard them 20:48:50 <tonyb> prometheanfire: infra will do the building for our internal mirrors 20:48:55 <smcginnis> Seems a little outside our scope. 20:49:04 <dirk> I guess I'm not following why we'd care about that 20:49:13 <dirk> do we want to avoid that the infra build-wheels job gets broken? 20:49:27 <prometheanfire> tonyb: ok, build only makes more sense 20:49:45 <tonyb> smcginnis: I see it as a completion of what check-uc* does 20:50:02 <prometheanfire> ya, co-installability extension 20:50:21 <tonyb> smcginnis: to be completely open I *thought* that check-uc did build wheels as part of the install process 20:52:09 <prometheanfire> I think it's worth doing, possibly in the same job as check-uc (since it has time and it's an extension) 20:52:11 <dirk> hmm, I thought I've seen bindep related issues there as well in the past 20:53:30 <tonyb> So it sounds like if I do some research and then write up the plan we can discuss it again noone is *hard* against it 20:53:43 <prometheanfire> sgtm 20:53:59 <prometheanfire> tonyb: so, three action items then 20:54:12 <prometheanfire> etherpad, pings and wheels 20:54:59 <tonyb> it could be a corner case where what we do in check-uc is fine but slightly diffeent to what LOCI needs/does in which case it might just be we add them as a cross job like nova or keystone 20:55:20 <prometheanfire> yep 20:55:22 <tonyb> huzzah 20:55:27 <prometheanfire> anyone have anything else? 20:55:32 <dirk> yeah 20:55:47 <dirk> I've seen the requests CVE bump abandoned, what did I miss again? 20:56:40 <prometheanfire> I didn't abandon it iirc 20:57:06 <prometheanfire> you have links to the reviews? abandon reason should be there 20:57:24 <dirk> https://review.opendev.org/#/c/640727/ 20:57:42 <prometheanfire> ugh 20:57:50 <prometheanfire> gertty was being stupid 20:57:56 <dirk> I am not sure I understand what "python is dumb" means 20:58:11 <prometheanfire> I was trying to abandon my 'test python intel' patch 20:58:16 <prometheanfire> no clue why it selected those 20:58:47 <dirk> I guess mistakes happen 20:58:54 <prometheanfire> ya 20:58:58 <dirk> can you unabandon it? I forgot to follow the mailthread, I'll do so 20:59:03 <prometheanfire> ya 20:59:07 <prometheanfire> just did 20:59:22 <dirk> there was some other ping that you sent me a week ago or so, but I wasn't able to reread it as I lost the irc history (was migrating bouncer implementation) 20:59:33 <dirk> what was the action item for me? 20:59:40 <tonyb> https://review.opendev.org/#/c/640710/ ? 20:59:41 <prometheanfire> so, I'm not sure we are any further for being able to do security patches 21:00:02 <prometheanfire> that one too, thanks 21:01:43 <prometheanfire> ok, all restored now (there were a couple of others) 21:01:55 <tonyb> okay 21:02:34 <tonyb> looks like https://review.opendev.org/#/q/topic:tox_constraints+is:open is moving along 21:02:40 <prometheanfire> anything else? 21:02:51 <tonyb> so I'll include that in the updates I do 21:03:00 <prometheanfire> thanks 21:03:43 <tonyb> dirk: do you mind if I add some Depends-On to you changes so we can things up safely? 21:04:52 <dirk> tonyb: the comment about django upper-constraints.. I think horizon installs unconstraint, so I guess changing upper-constraints is mostly a documentation exercise 21:05:14 <dirk> I'll add a link to the review, I wanted to double check things and maybe see if a fix can be done first 21:05:24 <tonyb> dirk: okay 21:05:26 <dirk> tonyb: you mean the tox_constraings part? yes, sure, go ahead 21:05:48 <dirk> tonyb: I wanted to lobby on -infra for a 2nd core review, but it seems people are distracted with some infra issue 21:06:05 <dirk> or rather, I didn't want to distract them from fixing the outage 21:06:21 <tonyb> dirk: yeah basially make sure we can't land things in a breaking order 21:06:35 <tonyb> dirk: cool beans 21:06:49 <tonyb> dirk: I can +1 the release update 21:08:49 <prometheanfire> ok, gonna end it here 21:08:57 <tonyb> +1 21:09:25 <prometheanfire> #endmeeting