[20:03:33] ok [20:03:39] time it is [20:03:51] who's chairing? ah right, me [20:04:21] * dilfridge has changed topic for #gentoo-council to: "Next meeting: 12 Nov 2013 at 19:00 UTC | http://www.gentoo.org/proj/en/council/utctolocal.html?time=19:00 | http://wiki.gentoo.org/wiki/Project:Council | Agenda: http://article.gmane.org/gmane.linux.gentoo.project/3149" [20:04:27] --> TomJBE (~tb@gentoo/developer/tomjbe) hat #gentoo-council betreten [20:05:00] blueness: dilfridge: rich0: scarabeus: ulm: WilliamH: Council meeting! [20:05:08] dberkholz: [20:05:16] -*- rich0 is here [20:05:23] 1 Roll Call [20:05:38] morning [20:05:39] -*- WilliamH here [20:05:44] morning! [20:06:11] -*- rich0 here [20:06:25] here here here [20:06:28] :) [20:07:00] blueness, ulm? [20:08:02] here [20:08:12] phew [20:08:21] -*- dilfridge was just searching for the phone number [20:08:33] i don't suppose anyone's put together a version of the agenda that embeds the emails that aren't in online archives? [20:08:48] anyone in the us wants to message blueness? [20:09:02] referring to things that are not trivial to access is less than friendly for people reading it [20:09:27] dberkholz: not yet but I should have most of them here, so if you all are willing to wait a moment [20:09:58] --> pacho2 (~pacho@gentoo/developer/pacho) hat #gentoo-council betreten [20:10:56] I'm finding it difficult to find those emails as well - does gmail support searching for msgids? [20:11:04] I do vaugely remember reading the originals... [20:11:48] --> Zero_Chaos (user18791@gentoo/developer/pentoo/zerochaos) hat #gentoo-council betreten [20:11:58] Could have been auto-deleted - I don't normally save list mail, figuring that it is supposed to be archived... :) [20:13:02] ok I copied [3,4,6], I doubt we get further [20:13:21] http://dev.gentoo.org/~dilfridge/mail-3.txt [20:13:24] http://dev.gentoo.org/~dilfridge/mail-4.txt [20:13:27] http://dev.gentoo.org/~dilfridge/mail-6.txt [20:13:52] I'll put something nicer together later [20:14:01] anyone got blueness phone number? [20:15:32] dilfridge: should I text him? [20:15:35] Ok I sent him a text [20:15:45] k [20:15:46] let's start now anyway, we're already late [20:16:06] Agenda topic 2: Status of the QA team [20:16:07] damnit! [20:16:12] i'm here the time change! [20:16:31] :) [20:16:35] i forgot about the time change to daylight savings so we start at 3 not 4 [20:16:37] start using the google calendar =) [20:16:43] yeah i know [20:16:44] heh [20:17:15] I've already voiced them, but I have concerns with our general approach to staffing QA (which is not meant to reflect on any team members, or actual team performance). [20:17:27] so basically we have two proposals related to QA... one by Polynomial-C, requesting to add Patrick; one by kensington, requesting dissolving and re-forming QA [20:17:48] It basically lacks oversight, is a thankless job, and seems to have considerable authority. [20:18:02] it would make sense to talk about kensington proposal first, since that one would make the other obsolete imho [20:18:12] okay [20:18:13] k [20:18:16] k [20:18:20] -*- WilliamH agrees [20:18:55] . [20:19:01] ulm: did you hear anything still? you're the qa guy here... [20:19:13] dissolving is pretty serious, we'd face the dicision of who do we staff QA with initially [20:19:25] dilfridge: diego has sent an answer to me privately [20:19:30] also, i'm not sure what diego's position is [20:19:36] blueness: it is, but the current staff is basically non-responsive. [20:19:47] ulm, can you share? [20:20:06] Certainly interested in knowing how the current QA staff feels. [20:20:16] blueness: unfortunately, the answer is of a nature that I'd rather not share it here [20:20:34] ulm, okay i can guess what he said [20:20:49] blueness: diego was asked a couple of times to get things going again... hwoarang was pretty active there [20:21:09] so [20:21:19] we need two things for QA, a team and someone willing to run the tinderbox stuff [20:21:19] also no replies to my call for devs interested to join qa: http://thread.gmane.org/gmane.linux.gentoo.project/3116/focus=3129 [20:21:21] if this is a route people want to go, how can we avoid repeating the same mistakes again with a new team? [20:21:27] I want to be sensitive to feelings, but if the current team can't even talk publicly about the problems faced by the current team, I think we need to rethink things. [20:21:34] i have somebody at the door, gimme sec to run over and see what is going on :) [20:21:39] rich0, ++ [20:21:51] what does QA actually mean for Gentoo? outside of things like repoman, etc? [20:22:02] seemant: defining that would be a good first step. :) [20:22:04] there's a whole glep about it [20:22:12] ulm: I don't see a call for interested devs. I'm interested, have been for years. [20:22:18] glep 48 defines qa [20:22:22] http://www.gentoo.org/proj/en/glep/glep-0048.html [20:22:33] The GLEP as written isn't bad. [20:22:34] Zero_Chaos: see the link I've posted above [20:22:44] "Could everyone who is interested in joining the QA team repeat their application here, so we can get a complete list?" [20:23:03] The only thing I'd change as I've stated before is that the QA lead should be confirmed by council - by all means let the team nominate somebody. [20:23:14] ulm: I'm not so worried about that, to be honest. It's easier to join a new team than to jump on a boat that's already rotting. [20:23:17] Zero_Chaos: maybe not everyone reads -project [20:23:28] ulm: I missed it in the thread, happy to take full blame but I might not be the only one that missed it. (I'm on -project) [20:23:29] I think if we re-constituted the team from scratch we'd probably get more takers. [20:23:41] rich0: exactly [20:23:46] back [20:24:03] is patrick interested? [20:24:06] I'm sorry, but the glep is unclear [20:24:11] I don't want being on the QA team to be a thankless job. [20:24:14] what is a "good state" for the portage tree? [20:24:32] blueness: I asked and he says he's basically neutreal to what we decide here, that's all I got [20:24:35] seemant: following http://devmanual.gentoo.org is a good start [20:24:42] anyway [20:24:52] I'm fine with QA setting and enforcing standards, but they should be written down. [20:24:54] I think we can all agree that the current state is not OK. [20:24:58] Sometimes they are. [20:25:27] dilfridge: Agreed. [20:25:33] dilfridge: ++ [20:25:44] well if the state of the current QA team is that bad, how about considering both recommendations [20:25:45] ulm: nobody wants to join QA as it is right now. that's why (i think) nobody replied [20:25:57] disolve it and put patrick in charge of reconstituting it [20:25:57] if the QA team has been dysfunctional/nonfunctional for this long, is it even something we need? [20:26:04] dberkholz, yes [20:26:10] we do [20:26:13] So, step 2A would be to vote on Kensington's proposal to dissolve QA and call for new team members. (we can talk later about the detailed proceedings for that.) [20:26:21] I'd like to first get the design of the team/purpose/etc worked out before we permanently staff it. [20:26:22] dberkholz: yes we do. [20:26:28] hwoarang: no one can join QA right now. multiple people (including me) claim to have tried and been ignored...as the bugs are ignored [20:26:29] looking at the bugs (and private threads) where qa@ is CC'd yes we need it [20:26:32] blueness: what problems are we currently having as a result of this? [20:26:38] the tinderbox runs are very useful even if they must be taken with a grain of salt [20:26:51] If patrick wants to help with that I'd be glad to have him on-board, but I think the council should approve the final result and confirm the new team. [20:27:03] dberkholz: comrel gets dragged into technical issues [20:27:08] So, you don't need a QA team to have a tinderbox. [20:27:09] dberkholz, its not a question of which problems, its a question of consistency between systems [20:27:10] i guess i'm asking whether it's rare enough that the council should directly resolve disputes [20:27:13] eg --as-needed [20:27:18] as an ldflag [20:27:29] Anybody can test anything and log bugs if they're helpful. [20:27:42] rich0, true [20:27:49] but i would like QA to also run the tinderbox [20:27:51] If Gentoo infra were being used I could see having a role to decide what it gets applied to/etc, but right now it is all private anyway. [20:28:01] or at least work closely [20:28:11] rich0: but not everyone has the authority to force changes [20:28:13] blueness: I'd love to have QA run the tinderbox, but Gentoo doesn't actually own a tinderbox. [20:28:15] rich0, we could make the tinderbox part of infra [20:28:26] blueness: of course [20:28:32] can't hurt to ask [20:28:39] Right now ComRel gets dragged into technical issues, which is not Comrel job. That is what the QA team is for, deciding what is appropriate for the tree. [20:29:04] I see QA as an executive team, and council as an oversight team. [20:29:05] ComRel can't slap anyone for committing garbage. [20:29:08] <-> johu_ heißt jetzt johu [20:29:15] dilfridge: the problem is that there aren't enough active members [20:29:20] We can set major policies, but that doesn't mean that we have to second-guess every little QA decision/policy/etc. [20:29:33] ulm: right now there are NO effective members. well, maybe luca. [20:29:38] I'd like to have the council confirm the QA team so that they have real backing and authority. [20:29:55] rich0: the current one? no way. [20:30:00] People would know that if QA speaks the council is likely to back them up. They could still appeal. [20:30:19] dilfridge: that's why I want to fix it first. I don't want to just rubber-stamp the status quo. [20:30:24] dilfridge: there are more active members [20:30:34] in subprojects mainly [20:30:37] i'd like to see the QA team be highly skilled programmers to watch for and even pre-empt issue and provide consistency across our packages wrt how they are compiled/built [20:31:01] ulm: we're not talking about subprojects, and kensington is also not talking about subprojects. [20:31:06] in my mind QA = police, council = judiciary [20:31:18] well we can do whatever we want but in result the current qa was quite demotivated by the results of most the appeals [20:31:19] I think we need to distinguish between the advisory side of QA (tinderbox, testing, etc), and the executive side of QA (power to force changes, ban, etc). [20:31:48] rich0, yes i agree with that, but I'd still like the two tied [20:31:48] scarabeus: that's why I'd like to make sure QA and Council are on the same page. That avoids all the second-guessing. [20:32:15] Sure, they are related, but ANYBODY can run a tinderbox and log bugs/etc. Nobody is going to object to that. [20:32:17] rich0: yeah, i think it's reasonable to give council additional oversight on QA and comrel vs other teams, given the authority they have over people outside their projects [20:32:33] What gets people fired up about QA is when somebody tells them they have to change something over their dead cvs access, or whatever. [20:32:41] rich0, someone needs to decide if a tinderbox result is something that is advisory or compulsory [20:32:41] yep [20:32:56] rich0: particularly to avoid them turning into insular "old boy" groups [20:33:10] Well, the results are advisory just due to their nature. It is the policy that they're running against that is compulsory. [20:33:20] rich0: +! [20:33:22] +1 [20:33:27] well, not replying at all to applications for new members is the best way to become an "old boys group" [20:33:30] +1 [20:33:42] Package ignores CFLAGS - that's information. Package isn't allowed to ignore CFLAGS - that's policy. [20:33:42] OK [20:33:51] NOW [20:34:10] shall we vote? [20:34:21] dilfridge, pose the question [20:35:04] 2A Kensington's proposal to dissolve QA and put out a call for new team members. (detailed proceedings for that to be decided afterwards.) [20:35:32] we should decide what to do [20:35:39] dilfridge, there are two questions there [20:35:44] not dissolve qa and then come up with a plan [20:35:50] ok [20:35:59] then let's discuss the proceedings first [20:36:18] here's a more detailed proposal (will take a moment to type) [20:36:31] I don't care whether we dissolve QA right now. What I'd like is to have a team start designing a "new QA." That doesn't need to be super-complicated. The goal would be to replace QA with the results of that, including initial staffing. [20:36:58] we could dissolve qa and have the council take the role of the qa lead temporarily, for accepting new members into the team [20:37:17] afterwards, qa would elect a new lead [20:37:23] (or even the old lead again) [20:37:37] --> graaff (~graaff@gentoo/developer/graaff) hat #gentoo-council betreten [20:37:43] ulm: honestly, I'd like to really get to a system where QA doesn't just run itself completely independently. [20:37:56] 2B. Dissolve QA (sans subprojects). Send an e-mail that interested devs should send Council a mail that they want to join QA. Council takes the role of the QA lead for two months, with elections in the team afterwards. Council accepts interested devs into the team with single secret majority vote. [20:38:14] rich0: then we need to change glep 48 [20:38:19] s/single// [20:38:32] ulm: Seems to be the season for changing GLEPs. :) [20:38:46] dilfridge, i don't know if i want this to be a completely democratic process though, i think the council should look for people with high technical skills [20:38:53] I'd suggest creating a next-gen QA team with the goal of designing a new GLEP 48. [20:38:53] i believe given the inactivity of the glep team, this will cause ever more delays... [20:39:04] dilfridge: sounds good [20:39:13] blueness: well it's up to us to decide. [20:39:23] okay [20:39:34] blueness: we're free to reject applicants [20:39:45] ok let me clarify the text once more, just a second [20:39:48] guess that's the reason for the secret vote [20:40:08] ulm, i'm just worried about who the lead will be, if QA has a good lead, then the rest will fall into place [20:40:19] 2C. Dissolve QA (sans subprojects). Send an e-mail that interested devs should send Council a mail that they want to join QA. Council takes the role of the QA lead for two months, with elections in the team afterwards. Council accepts interested devs into the team with secret majority vote of the council members. [20:40:50] What do you mean "sans subprojects"? [20:40:54] dilfridge: e-mail to whom? -dev-announce? [20:41:06] I mean that we're not dissolving e.g. treecleaners. [20:41:17] e-mail to council. [20:41:26] sounds good [20:41:41] dilfridge: no, I meant where we would send the e-mail ;) [20:41:57] ulm: council@g.o [20:42:09] we send our mail to gentoo-dev-announce to get maximum coverage. after all that's where all devs must be subscribed. [20:42:11] 2C sounds good [20:42:18] k [20:42:21] we accept applications at council@gentoo.org [20:42:44] Do applications have to be secret? [20:43:08] I dont think so, you can always copy the mail to somewhere else [20:43:14] (if you want) [20:43:24] Arfrever: well, we wouldn't stop people from saying that they have applied if they wanted to; that's up to them. [20:43:39] I'd really like the proposed lead to be aligned with the council. I'm still not sure that it is a good idea to just pick somebody and have them take over with the current structure. [20:43:40] what I want to avoid is, [20:43:55] If the current QA team feels like the council undercuts them, what will stop the new team from feeling this way? [20:43:59] that someone is turned down and this becomes some sort of public shaming [20:44:10] rich0, the current QA team can speak for themselves [20:44:27] dilfridge, correct, the point of secrecy is to avoid hurt feelings [20:44:57] anyway, the new QA lead in two months will be able to accept members too [20:45:00] dilfridge: two months seems a bit too long [20:45:01] blueness: I just don't think that the self-governing but powerful team with no formal oversight other than appeal system is really a good design. We run all our big teams this way, and I think it can make them insular. [20:45:10] i would like to see it move toward a model where the QA team votes for a lead but it requires council approval to put them in place. [20:45:15] I think nominations should only be open for a week or two. [20:45:29] Then I think we really should try to interview or otherwise interact with the candidates. [20:45:36] dberkholz: ++ [20:45:49] dberkholz: good plan, how do we implemenet it? [20:45:54] dberkholz, yes eventually [20:46:05] Well, we can always just make it a policy. [20:46:13] Or amend the current GLEP to say that. [20:46:30] Just change the second bullet and leave the rest alone to start. [20:46:31] dberkholz: that's against the systematics in all other projects [20:46:37] shall we make "amedning the glep to say that" an additional vote? [20:46:55] ulm: I think QA and comrel are not ordinary projects. [20:47:00] ulm: yeah but qa is a special project (as is comrel) [20:47:07] ulm: QA and comrel are exceptional projects because they have power over developers not part of their project [20:47:08] and I wouldn't want to be a member of a project with a lead not appointed by the project members themselves [20:47:15] If QA were like all other projects, then I could start my own QA project today and be the lead of that project. [20:47:39] Other projects don't have distro-wide authority. [20:47:42] ulm: elected by the members, confirmed by the council? [20:48:02] (not, chosen by the council but council can reject) [20:48:08] s/,// [20:48:13] if we don't trust the new qa to be able to elect a reasonable person as team lead, then we should just give up [20:48:15] If the council doesn't confirm, they can pick a new lead. If things aren't worked out in some period of time council appoints a lead. [20:48:31] ulm: giving up on the current system is EXACTLY what I'm proposing. [20:48:32] anyway this all is a separate discussion [20:48:45] Why not just have each council appoint the new council. Can't we be trusted to do a good job with that? [20:48:53] ulm: isn't that what we've already done just by having this item on the agenda? [20:48:53] let me re-formulate the proposal one last time... [20:48:56] rich0, heh [20:49:01] ulm: it seems like you're arguing that the current situation could never happen again [20:49:27] we have it on the agenda because sort of a deadlock sitation arose [20:49:34] *situation [20:49:39] the argument might go like this: the council is elected by all devs, so we are being trusted with judiciary powers. QA is not elected by all devs and yet has power over all devs. so it needs extra oversight [20:49:48] that doesn't mean that it will happen again [20:50:03] that extra oversight is the council approves the lead, to make sure the QA team doesn't go crayz [20:50:21] blueness: that, and the QA team has some kind of mandate [20:50:30] blueness: as I said, glep 48 needs to be changed then [20:50:40] ulm, i agree i think it does [20:50:43] Well, changing GLEP 48 just to codify that is simple. [20:50:48] 2D. Dissolve QA (sans subprojects). Send an e-mail to gentoo-dev-announce that interested developers should send council@gentoo.org a mail in the next two weeks that they want to join QA. Council takes the role of the QA lead for six weeks, with elections in the team afterwards. Council accepts interested devs into the team with secret majority vote of the council members. [20:50:53] Let the team work out the details. [20:51:00] The other side of extra oversite is that there is no point to appealing to the council if qa does something you don't agree with? [20:51:00] Let us vote on 2D now and discuss the GLEP afterwards. [20:51:18] yes, glep 48 needs about 5 words added to the 2nd bullet "and confirmed by the council" [20:51:30] I'm fine with voting on that, as long as the council is under no obligation to appoint anybody in six weeks. [20:51:34] i'm ready to vote [20:51:49] -*- dilfridge yes on 2D [20:51:55] -*- rich0 yes on 2d [20:51:56] -*- blueness yes on 2D [20:52:10] yup [20:52:16] please restate 2d again? It has scrolled off? [20:52:24] 2D. Dissolve QA (sans subprojects). Send an e-mail to gentoo-dev-announce that interested developers should send council@gentoo.org a mail in the next two weeks that they want to join QA. Council takes the role of the QA lead for six weeks, with elections in the team afterwards. Council accepts interested devs into the team with secret majority vote of the council members. [20:52:53] I abstain [20:52:55] as I am a member of the current QA team [20:53:49] dberkholz: WilliamH? [20:54:21] did we loose them? [20:54:39] -*- WilliamH is still here thinking... [20:54:42] abstain, it already passed and i'm not confident enough that it's the right answer [20:55:10] abstain also; I'm not confident. [20:55:15] dberkholz: I plan to offer some proposals on a revised GLEP 48 - there is no reason that we can't pass that before we appoint a new QA lead. [20:55:27] ok that's 4 yes and 3 abstain, motion passed. [20:55:29] No reason we can't get the ball rolling though. [20:55:40] Now about GLEP48 [20:55:58] rich0: I think the new qa team should at least be heard on their opinion on a new glep 48 [20:56:11] -*- WilliamH tends to agree [20:56:12] Sure, but I'll hear that before I vote to put anybody on a new QA team. [20:56:19] even if it's up to the council to approve it [20:56:24] yeah we'll need to send the mods around for discussion [20:56:24] There won't be a new QA team as far as I'm concerned until those issues are sorted out. [20:56:50] However, no need to bikeshed it right now. [20:56:55] ulm: we can change the rules later again, but for now we need some set of rules to play by [20:56:59] rich0: I think it is fair to ask the "new QA team" to fix the glep as their first official act. [20:57:10] rich0: My concern about making the lead chosen by the council is that can get into a situation where [20:57:11] OK let's vote on [20:57:18] I'm also fine with interim appointments - such as giving them a six month term to fix things/etc. [20:57:19] i think we should add "and confirmed by the council" to the 2nd bullet point, 1st sentence, as a draft for discussion [20:57:21] 2E should we change GLEP48 now? [20:57:26] and send it around to -dev then vote [20:57:55] dberkholz: agree. I can send a draft out to the list, and we can vote next meeting? [20:58:00] dilfridge, 2E is too open, sould we say what we're changing? [20:58:04] or -project, wherever seems appropriate [20:58:13] OK [20:58:20] the voting next time sounds feasible, and we should sent this to project :) [20:58:26] probably -project [20:58:34] certainly a unilateral vote on a glep change by council without any opportunity for community discussion seems wrong [20:58:43] -*- WilliamH agrees with dberkholz [20:58:56] dberkholz: scarabeus: next meeting sounds good, since that is also still before the qa team election [20:59:00] Well, not bad - we got through half of one agenda item in an hour. :) [20:59:09] That was a big one though. [20:59:10] How long are we taking applications for members? [20:59:16] two weeks [20:59:18] Per the proposal - two weeks. [20:59:28] ok [20:59:32] okay with me [20:59:48] I think Polynomial-C's proposal is now off the table, does anyone disagree with that? [20:59:58] :) [20:59:58] (adding Patrick to the QA team) [21:00:16] If he wants to be added he can feel free to apply. :-) [21:00:18] good [21:00:45] now since we've reached the magical limit of 60min, let's stop here and reconvene in a week... [21:00:59] heh ok. [21:01:00] and I'll summarize the result [21:01:04] okay [21:01:05] hehe oky [21:01:12] that was a thorny issue though [21:01:30] well yes, we definitely decided something. [21:01:38] -*- dilfridge bangs on the table. [21:01:40] It needed to be addressed; the qa situation wasn't good. [21:01:42] meeting closed. [21:02:12] :)