| *** hieulq has quit IRC | 01:00 | |
| *** namnh has joined #openstack-tc | 03:22 | |
| namnh | EmilienM, hi Emilien, For your comment on this patch: https://review.openstack.org/#/c/472547/ . Our the grenade gate was moved to voting, so could you please review the patch again. Thanks so much. | 03:36 |
|---|---|---|
| namnh | EmilienM, here is the patch to do this: https://review.openstack.org/#/c/472153/ | 03:36 |
| *** mugsie has quit IRC | 07:51 | |
| *** mugsie has joined #openstack-tc | 07:54 | |
| *** mugsie has quit IRC | 07:54 | |
| *** mugsie has joined #openstack-tc | 07:54 | |
| *** jpich has joined #openstack-tc | 08:11 | |
| *** mugsie has quit IRC | 08:59 | |
| *** mugsie has joined #openstack-tc | 09:00 | |
| *** mugsie has quit IRC | 09:00 | |
| *** mugsie has joined #openstack-tc | 09:00 | |
| * ttx quietly drops a few bombs to the ML | 09:15 | |
| *** mugsie has quit IRC | 09:22 | |
| *** mugsie has joined #openstack-tc | 09:22 | |
| *** mugsie has quit IRC | 09:22 | |
| *** mugsie has joined #openstack-tc | 09:22 | |
| *** mugsie has quit IRC | 10:27 | |
| *** mugsie has joined #openstack-tc | 10:29 | |
| *** mugsie has quit IRC | 10:29 | |
| *** mugsie has joined #openstack-tc | 10:29 | |
| *** cdent has joined #openstack-tc | 10:33 | |
| *** sdague has joined #openstack-tc | 10:48 | |
| EmilienM | namnh: ok, I'll look today | 11:11 |
| EmilienM | namnh: I commented https://review.openstack.org/#/c/472153/3/zuul/layout.yaml | 11:12 |
| EmilienM | namnh: +1 on https://review.openstack.org/#/c/472547/ - but please address my comment for the zuul layout | 11:15 |
| namnh | EmilienM, I see, thanks for your +1. I will push up a patch set to update the gate. | 12:09 |
| *** namnh has quit IRC | 12:10 | |
| EmilienM | cool | 12:14 |
| cdent | opium++ | 13:00 |
| *** mugsie has quit IRC | 13:43 | |
| *** mugsie has joined #openstack-tc | 13:46 | |
| *** cdent_ has joined #openstack-tc | 14:11 | |
| *** cdent has quit IRC | 14:11 | |
| *** cdent_ is now known as cdent | 14:11 | |
| smcginnis | cdent: Heroine? :) | 14:12 |
| cdent | heh | 14:12 |
| cdent | I'm struggling to explain to myself why infra should host everyone and everything that wants to be hosted | 14:14 |
| smcginnis | I had the same thought. | 14:15 |
| cdent | and opium kind of ends up being a good word for that level of dedication | 14:15 |
| smcginnis | Hah! | 14:15 |
| cdent | as far as I can tell there isn't even agreement with the tc on what "big tent" means | 14:28 |
| cdent | so at least killing that is a start | 14:28 |
| cdent | but unless we answer the questions, the next set of stuff will have the same issues | 14:28 |
| ttx | cdent: if we had a clear name to describe "what is hosted on infra but actually not officially openstack", I posit that we would have avoided the confusion around what "big tent" means | 14:31 |
| cdent | ttx: probably true, but we'd still have the confusion over what "officially openstack" means :) | 14:32 |
| cdent | (or, rather, why it matters) | 14:32 |
| ttx | In the end, it really means "governed by the Technical Committee" | 14:32 |
| cdent | which is a non-statement | 14:33 |
| cdent | it doesn't say anything about value | 14:33 |
| ttx | since being in gives you voting rights, and places you under the TC oversight in return | 14:33 |
| *** emagana has joined #openstack-tc | 14:33 | |
| ttx | well, to use the trademark, you have to be governed by the TC | 14:33 |
| ttx | so the value is all about the ability to say that you are "an OpenStack project" | 14:33 |
| cdent | but you don't get the trademark just from that, do you? | 14:34 |
| ttx | side benefits (like event space) derives from that | 14:34 |
| cdent | you still have to pass the tests, etc | 14:34 |
| ttx | cdent: for some of trademark usage, being official is enough | 14:34 |
| ttx | like being able to call yourself an OpenStack project | 14:34 |
| cdent | we seem to be in a tautology | 14:35 |
| ttx | called community usage of the trademark | 14:35 |
| ttx | basically the TC defines the limit of what can use the openstack name and what's not | 14:35 |
| ttx | or rather, we define the limits of the community, which in turn defines which deliverables can call themselves openstack | 14:35 |
| cdent | Yeah, I know all that, but I think if we want to really get this stuff right, eventually, we need to be able to enumerate (at least to ourselves) why anyone (from a project) should care about any of that. | 14:36 |
| cdent | simply being able to say "I'm a part of OpenStack" is nice, but it doesn't say enough | 14:36 |
| ttx | I covered some of it in https://ttx.re/create-official-openstack-project.html | 14:36 |
| * cdent remembers that | 14:36 | |
| ttx | At this point I think there are two issues we need to address... One is the fact that some people think "big tent" means "unregulated". The other is that a lot of projects nourish the confusion by saying they are an openstack project, when they are just merely hosted | 14:38 |
| ttx | 1st part is easy, just remove the confusing "big tent" brand | 14:38 |
| ttx | Second part is harder, you kinda need to give that "other state" a name | 14:38 |
| ttx | "unofficial openstack project" just doesn't cut it | 14:39 |
| cdent | So I think the latter one is where we get it wrong. If you're using openstack infra (lists, irc, logging, CI) then you are an openstack project, you are part of the openstack community. Saying anything else is just mean. | 14:39 |
| cdent | the thing that needs to have its name changed are not the entire suite of openstack projects, but those projects which are "governed" | 14:39 |
| ttx | cdent: currently we are saying that to be part of the openstack community, you need to have alignment on values, principles, tooling and mission. | 14:40 |
| ttx | Being on infra just checks for tooling | 14:40 |
| ttx | For the rest, you apply to the TC | 14:40 |
| cdent | we _say_ that, but from the outside that's now how it looks | 14:40 |
| cdent | if I can find code for which the git url includes openstack... | 14:40 |
| cdent | anyone would conclude: this code _is_ openstack | 14:41 |
| ttx | fwiw I'm campaigning for removal of all prefixes | 14:41 |
| *** mugsie has quit IRC | 14:41 | |
| ttx | cdent: agree that doesn't help | 14:41 |
| cdent | how do you imagine removing prefixes to work with mirroring to github? | 14:41 |
| ttx | especially since we are doing it for technical reasons (pain of renaming) rather than for branding reasons | 14:42 |
| ttx | cdent: you would still mirror to openstack/* anything that's on OpenStack's gerrit ? | 14:42 |
| *** mugsie has joined #openstack-tc | 14:42 | |
| *** mugsie has quit IRC | 14:42 | |
| *** mugsie has joined #openstack-tc | 14:42 | |
| ttx | and make the org be about a mirror of everythign you can find there ? | 14:42 |
| cdent | people will still conclude: these repos are openstack | 14:43 |
| ttx | cdent: who uses GitHub anyway ? Oh wait | 14:43 |
| cdent | heh | 14:43 |
| ttx | Another tactic would be to reinstate a prefix, like stackforge/ | 14:43 |
| cdent | I'm not trying to be difficult here, just point out one of the many different angles... | 14:43 |
| ttx | we have a lot less movement between the two than we used to | 14:44 |
| ttx | or opium/ | 14:44 |
| ttx | cdent: heh, I know. If the solution was simple, we would have fixed that a long time ago | 14:45 |
| ttx | I just fear that just removing all occurences of "big tent" in the doc won't be sufficient | 14:45 |
| ttx | unless we give a new catchy name | 14:45 |
| cdent | I have that fear too, thus the questions I posted | 14:45 |
| sdague | I wonder how hard it would be to change the mirroring so that there were no prefixes in gerrit, but there were when mirrored to github | 14:45 |
| sdague | and if that would help with the issue | 14:46 |
| ttx | which is why I combined the two things. Find a new name for the category, and explain, we don't say "big tent" anymore, we say X | 14:46 |
| ttx | sdague: true, we could still mirror to two different orgs | 14:46 |
| ttx | based on projects.yaml | 14:46 |
| sdague | also, is there a reason to not have this conversation over in #dev? this impacts more folks and it's not clear that this needs to be a tc only input space | 14:47 |
| ttx | sdague: nope, I was just anticipating office hours | 14:47 |
| ttx | Office hours now! | 15:01 |
| johnthetubaguy | yeah, cdent, I promised I would re-read that vision draft | 15:02 |
| cdent | I don't reckon there's any huge rush? | 15:02 |
| cdent | It won't be 2019 for a while yet | 15:02 |
| johnthetubaguy | oh dear... | 15:03 |
| johnthetubaguy | there are no subtitles added yet, was that on purpose? | 15:03 |
| johnthetubaguy | in some ways it flows too well right now to add them | 15:04 |
| cdent | johnthetubaguy: not really, I just didn't do that part yet as I was in the mode of "see how the paragraph transitions work out" | 15:04 |
| cdent | but where they should go is pretty clear | 15:04 |
| fungi | this seems way more active than the last office hours | 15:05 |
| ttx | fungi: indeed | 15:05 |
| fungi | (which was me, all me, all the time) | 15:05 |
| fungi | got bored talking to myself pretty quickly | 15:06 |
| ttx | fungi: replied to your email. I see what you mean by not needing to define the negative term. Trick is, without a term, there are enough confusing signs (like the "openstack/" prefix in the git repo name) to imply that everything is an official openstack project | 15:06 |
| cdent | johnthetubaguy: if you're inclined to either leave a comment reminding me, or just adding them yourself, both are great with me. I'm going to be funemployed for the next two weeks but still attending to that kind of stuff. | 15:06 |
| ttx | It's actually the combination of the two that's killing us :) So we need to fix one or the other, or both | 15:07 |
| ttx | Currently, "unofficial openstack project" is still seen as "openstack project" by most | 15:08 |
| fungi | making gerrit mirror to different github prefixes on a per-repo basis would get messy quickly... like adding thousands of lines to its configuration and hoping it doesn't explode | 15:08 |
| ttx | so the negative space approach is not working | 15:08 |
| sdague | negative space only works if you are really familiar with the space in question | 15:08 |
| ttx | sdague: ++ | 15:08 |
| johnthetubaguy | cdent: cool, I see your TODO, I guess I can comment on that. | 15:08 |
| sdague | we're "mostly" not clarifying that part | 15:09 |
| sdague | this is about clarification for people that aren't super familiar | 15:09 |
| fungi | my main concern with "branding" anything is that the next step is always "promoting" | 15:09 |
| ttx | sdague: right, and I really think giving it a name will help | 15:09 |
| ttx | fungi: we didn't see that much promotion in the stackforge days | 15:09 |
| sdague | fungi: what promotion are you concerned about? | 15:09 |
| fungi | except we did. for example people thought you needed to "join stackforge" before you could become incubated | 15:10 |
| ttx | sdague: projects clamoring that they are an opium project ? | 15:10 |
| fungi | and people always talked about "incubating in stackforge" | 15:10 |
| mugsie | fungi: did you not have to join stackforge to become incubated? | 15:10 |
| fungi | we let people co-locate their software development with openstack as a measure of convenience, mainly because the board said companies wanted to be able to do that | 15:10 |
| mugsie | one fo the requirements for incubation was to use the infra toolds | 15:10 |
| fungi | mugsie: you very much did not need to, no | 15:10 |
| fungi | mugsie: yes, you could go from somewhere else to incubated. there wasn't a required stackforfge-first step | 15:11 |
| mugsie | did anyone ever? | 15:11 |
| ttx | fungi: which is why the name should not imply incubation | 15:11 |
| ttx | stackforge was a bit bad in that respect | 15:11 |
| ttx | "hosted" projects would not have that connotation | 15:12 |
| ttx | "opium", maybe | 15:12 |
| fungi | i'm starting to like the term co-located | 15:12 |
| notmyname | ttx: based on what you said above, seems that one big concern about "what is openstack" is related to the *deliverable* not the hosting. that is, kill any notion of "openstack projects" and instead just focus on "openstack deliverables"? ie the previously-named integrated release | 15:13 |
| notmyname | to be more clear, "openstack is the released thing" and do away with the idea of "an openstack project" | 15:13 |
| ttx | notmyname: it's tricky, because some of the deliverables produced by official projects are not openstack deliverables. Think infra tools | 15:13 |
| fungi | i still call them "projects" when i mean "deliverables" so that seems fine to me | 15:13 |
| fungi | because all our tooling calls each git repo a "project" | 15:14 |
| sdague | fungi: are we mirroring directly to github from gerrit? or is it a secondary mirror from git.openstack.org? | 15:14 |
| ttx | Also we tried to kill "projects" before, but that didn't succeed :) | 15:14 |
| fungi | sdague: we're using gerrit's replication plugin | 15:14 |
| notmyname | ttx: no, there wouldn't be such things as an "official project" at all! | 15:14 |
| ttx | (currently we should only say "project teams" and "deliverables") | 15:14 |
| ttx | notmyname: I would argue that's already the case (for last two years), and yet people still say "openstack projects" | 15:15 |
| ttx | as a way to describe a group of deliverables I guess | 15:15 |
| ttx | Like... glance and glancestore are the glance project | 15:16 |
| ttx | notmyname: we hoped that releases.openstack.org would be seen as the canonical truth of what is openstack, but apparently people consider github pages as more canonical | 15:17 |
| * fungi would still love to drop github mirroring entirely | 15:18 | |
| fungi | i suppose we could rename the github org for openstack to "openstack-and-pals" | 15:18 |
| notmyname | well, if you take the thought about .5 steps farther, then you just say "openstack is a distro of cloud software" (regardless of where it's hosted/maintained). but that puts us in a place pretty far away from where we started originally | 15:18 |
| fungi | to imply that it hosts more than just openstack projects | 15:18 |
| cdent | fungi: I think the problem is that they _are_ openstack projects. using openstack infra is what makes them so. not names or governance policies. that we say they are not is pretty much meaningless. | 15:19 |
| cdent | so we should change the name of the official projects, not the unofficial ones | 15:19 |
| sdague | I would be very sad if github mirroring was dropped, given that it's the only way to be able to provide a stable url link to a specific commit that includes highlighted lines | 15:20 |
| ttx | We basically removed all friction when a project joins (no change in ML, IRC channels, repository names), at the price of creating a lot of confusion in the marketplacefungi: I wonder if the on-boarding convenience | 15:20 |
| lbragstad | sdague: ++ | 15:20 |
| ttx | oops | 15:20 |
| ttx | We basically removed all friction when a project joins (no change in ML, IRC channels, repository names), at the price of creating a lot of confusion in the marketplace | 15:20 |
| ttx | I wonder if that's not too much of a price to pay for on-boarding convenience | 15:21 |
| cdent | (highlighted lines)++ | 15:21 |
| ttx | i wouldn't mind if things that are not official would use separate MLs and require renaming of git repo when they get in. It's not as if we converted that many recently | 15:21 |
| fungi | i'm also fully in favor of declaring stackforge a failed experiment and asking remaining projects to either join openstack officially or relocate within some timeframe | 15:21 |
| ttx | fungi: that would work | 15:22 |
| cdent | fungi: that's an option that is certainly worth considering, although it somehow feels mean for some reason | 15:22 |
| fungi | would solve all of the ambiguity right there | 15:22 |
| fungi | stop hosting things that aren't openstack | 15:22 |
| mugsie | what about drivers for things like neutron, that cannot join openstack | 15:22 |
| ttx | fungi: totally. | 15:22 |
| cdent | I think it would help clarify (in the sense of "distill") openstack | 15:22 |
| mugsie | but still get used by openstack installs | 15:22 |
| fungi | mugsie: there are a number of them whose vendors already host them elsewhere | 15:23 |
| ttx | mugsie: would force us to clarify their status | 15:23 |
| ttx | which wouldn't be a bad thing | 15:23 |
| ttx | currently there is a lot riding on that grey negative space | 15:24 |
| mugsie | yeah | 15:24 |
| ttx | either being hurt of exploiting that confusion | 15:24 |
| ttx | or* | 15:24 |
| johnthetubaguy | do we have a good point in the thread with a consolidated list of issues we are trying to fix? | 15:24 |
| ttx | fungi: I agree that the best way to clear it would be to get rid of it | 15:24 |
| ttx | The projects (see what I did here) that would hurt the most would be those that are clearly not in our scope but still kinda want to live here | 15:26 |
| ttx | fungi: how many % resources would you say would you say stackforge^Wopium^Wnegative space takes those days ? | 15:27 |
| sdague | so... one of the reasons we had this "other" space was because projects that are growing up from new ideas have some place to work like an openstack project from quite early (possibly commit 1) | 15:28 |
| ttx | fungi: having a bit of data on that negative space (how many repos, what activity, how many added recently...) would help assess the cost of getting rid of it | 15:28 |
| sdague | because the way openstack review tools and testing works is quite different than in the rest of the world | 15:28 |
| notmyname | sdague: maybe that's a bug and not a feature? | 15:29 |
| mugsie | i think it is a feautre | 15:29 |
| sdague | notmyname: maybe. Though you really can't do the kind of testing we do on travis | 15:29 |
| sdague | and having participated in code review in github for large projects.... oof is is brutal vs. gerrit | 15:30 |
| fungi | ttx: i don't think we have any real numbers around that (and would depend a lot on what we want to measure). i'll see what i can come up with | 15:30 |
| ttx | sdague: if we want to keep that "training ground" that points to the need for a name for it. Plus rules for culling it when things don't go anywhere | 15:31 |
| sdague | ttx: right | 15:32 |
| sdague | just trying to explore what's important to people | 15:32 |
| *** gordc has joined #openstack-tc | 15:33 | |
| *** tdasilva has joined #openstack-tc | 15:33 | |
| cdent | to continue the exploration: is the need for a training ground an indication of a bug? | 15:33 |
| cdent | (on the other hand, I very much like that we have been in the position to be able to provide such thing...) | 15:34 |
| ttx | sdague: it's a fair point. I think those are two solutions that would solve the problem. The Opium route (giving the negative space a name), or removing the space, or a mix of the two (have opium with strict rules about projects needing to move one way or another within x months) | 15:34 |
| cdent | (it's a huge luxury/privilege) | 15:34 |
| ttx | cdent: it's not that much a training ground imho. It's more a way to not force people to use github to land the first commits | 15:35 |
| ttx | although I do that | 15:35 |
| fungi | or renaming the org on github to be more obviously not "just openstack" (while moving forward with getting rid of namespaces in our gerrit) | 15:35 |
| sdague | ttx: I disagree, it really is a training ground | 15:35 |
| sdague | gerrit is something new to lots of folks | 15:35 |
| notmyname | cdent: the need for a "training ground" is more around what you said in your email: the need for marketing. that's not always a bad thing; openstack has been a place where competitors can legally cooperate, so having an openstack project enables that (whereas a separate namespace is a barrier because of the extra legal scrutiny) | 15:36 |
| ttx | sdague: do people in the negative space actually use that much of our unique approadch, beyond having to use Gerrit ? | 15:36 |
| ttx | i.e. how many projects actually use real tests | 15:36 |
| sdague | ttx: that's a good question | 15:36 |
| sdague | which has a real answer that could be found | 15:36 |
| ttx | right, looks like choosing the right approach will call for a bit of demographics | 15:37 |
| ttx | I'm not attached to a particular solution. I just don't think the status quo is bearable anymore | 15:38 |
| ttx | Been talking more with people outside of our usual circle, and the confusion is really gigantic out there | 15:38 |
| ttx | and tbh I can't blame them | 15:39 |
| sdague | I do think that if projects grow up in travis world, and have to invest a bunch of time in bug-bots to manage github issues, it probably means too much investment to come over to any system we end up having | 15:39 |
| sdague | ttx: I agree on the confusion part, definitely happy to make a change there | 15:39 |
| ttx | On another note, if anyone wants to discuss the Fuel proposal, I'm here! | 15:40 |
| smcginnis | ttx: It would be great to get more of a voice from those outside, since that is where most of the confusion happens. | 15:40 |
| notmyname | sdague: I want to go back to your comment about how openstack's testing is different than "the rest of the world" | 15:40 |
| sdague | notmyname: sure | 15:40 |
| smcginnis | ttx: Maybe that should go over to the operators mailing list and ask for input? | 15:40 |
| sdague | notmyname: where I'm defining "rest of the world" as what you'd find out on github (so it's a narrower slice of the rest of the world) | 15:41 |
| sdague | but where I think the alternative mostly will be for people not in our infrastructure | 15:41 |
| notmyname | sdague: my response (maybe that's a bug and not a feature) isn't a value judgement on either. what i mean is that the differences and the perception is what's adding to the confusion of "what is openstack" | 15:42 |
| ttx | smcginnis: This proposal of mine is actually the result of work we have been doing on the "communicating what is openstack" workgroup from the TC+Board+UC workshop | 15:42 |
| notmyname | sdague: it get's back to the questions cdent was asking about why people even think the naming question is important | 15:42 |
| ttx | smcginnis: it's not coming out of the blue | 15:42 |
| notmyname | sdague: *why* do people want to have a name for half-in-but-not-really openstack projects? | 15:43 |
| ttx | smcginnis: I can probably get lsell to comment on that thread | 15:43 |
| notmyname | sdague: what value to people get by being "openstack"? | 15:43 |
| smcginnis | ttx: +1 | 15:43 |
| sdague | notmyname: because without a label for it, everyone not deeply knowledgable in our community assumes every single thing on github.com/openstack is openstack | 15:44 |
| sdague | and like sourceforge, that include a lot of half baked rot | 15:44 |
| notmyname | sdague: for all practical purposes, it is! (today!) | 15:44 |
| sdague | notmyname: except it isn't, that was never the intent | 15:45 |
| notmyname | it doesn't matter what the intent is | 15:45 |
| fungi | the impression i get is that it's, say, the stereotypical clueless newb whose manager tells them they need to "install an openstack" and now they're stuck trying to figure out what that even means | 15:45 |
| notmyname | if it's under that URL, it's an openstack project | 15:45 |
| ttx | smcginnis: the most confused are the press (but then they are easily confused), followed by analysts, followed by CxOs and random architects. Not sure asking our MLs would reach the target audience. But Lauren talks to them all :) | 15:45 |
| smcginnis | Cool. Just thinking if we just discuss within this same group, we'll just end up in another tent. | 15:46 |
| fungi | and if we can't make it easy for that engineer who needs "an openstack" to discover what's part of openstack and what isn't, then there's a problem | 15:46 |
| cdent | smcginnis++ | 15:47 |
| ttx | smcginnis: yep. Which is why the discussion started at that Board+TC+UC group, with heavy input from marketing :) | 15:47 |
| ttx | It's no longer a technical issue | 15:48 |
| ttx | But it's still a TC one | 15:48 |
| smcginnis | Yep, get that. | 15:48 |
| sdague | notmyname: right, I think that's one of the key problems :) | 15:48 |
| sdague | https://github.com/openstack/bareon-api is openstack (as an example) | 15:49 |
| ttx | sdague: and if you read that "openstack is a big tent" looking at that github org kinda confirms your worse fears about what that term means | 15:49 |
| sdague | right | 15:49 |
| ttx | so the combination of the two is hurting us a lot | 15:50 |
| * johnthetubaguy nods | 15:50 | |
| * smcginnis nods too | 15:50 | |
| sdague | at some level perception is reality, as people are making decisions based on it about whether or not they get engaged | 15:50 |
| notmyname | yes! | 15:51 |
| johnthetubaguy | yup | 15:51 |
| smcginnis | And I can say, now working for a vendor that has a downstream distro, decisions are made based on that more than I realized. | 15:51 |
| ttx | Also lately, was used by... let's say... alternative organizations... as a cheap way to ditch openstack | 15:51 |
| johnthetubaguy | smcginnis: I believe, becuase people buy stuff that way too | 15:51 |
| smcginnis | johnthetubaguy: +1 | 15:51 |
| notmyname | so when we (the internal openstack community) figure out why people want to have a project with the openstack prefix, then we can better understand what to do to reduce confusion | 15:52 |
| johnthetubaguy | this seems to be where we all started, but how many people would it hurt to move repos not in projects.yaml to openstack-universe or something? | 15:55 |
| johnthetubaguy | ... so I think I am just understanding the through process you were all going through about 55 mins ago | 15:56 |
| johnthetubaguy | giving it a name, vs naming the official stuff better (constellations), etc. | 15:57 |
| johnthetubaguy | s/through/thought/ | 15:57 |
| notmyname | ok, crazy proposal (if only to explore the limits of the solution space): what if we killed the whole idea of "openstack projects" and focused only on the deliverables? so cinder is a project that does its own thing, and it's included in the openstack distro. and maybe also the k8s distro. and maybe something else (substitute for any other current openstack project or hypothetical release packag | 16:00 |
| notmyname | e) | 16:00 |
| smcginnis | notmyname: So we each get our own tent. | 16:01 |
| notmyname | ttx: not killing "projects" as was attempted earlier, but more like the stuff that's in -infra is the *only* "openstack" there is | 16:01 |
| notmyname | smcginnis: yeah | 16:01 |
| notmyname | smcginnis: or hut or yurt or house or whatever you want to build, actually | 16:02 |
| * smcginnis wants a hammock | 16:02 | |
| smcginnis | notmyname: Interesting idea. | 16:02 |
| notmyname | the implication would be that "openstack" would likely directly compete with RHEL and ubuntu | 16:03 |
| notmyname | and there's definitely loss around the internal community | 16:03 |
| cdent | so "building openstack" is the sort of mission? | 16:03 |
| cdent | and there are things that feed that | 16:03 |
| notmyname | cdent: yes | 16:04 |
| notmyname | you could even have "openstack" that includes k8s. or whatever else is the hotness of the month | 16:04 |
| ttx | notmyname: I think that is a bit far reaching, and a bit orthogonal with the problem of the negative space creating cinfusion | 16:05 |
| notmyname | (I'm spitballing this whole thing at the moment) it would be a change to focus on what's delivered vs the current pattern of *how* it's delivered | 16:05 |
| notmyname | ttx: oh, it's definitely far reaching! like I said, only trying to find the boundaries of the solution space :-) | 16:06 |
| *** jpich has quit IRC | 16:06 | |
| ttx | notmyname: also how much different is it from sayign openstack = what's on releases.openstack.org ? | 16:06 |
| johnthetubaguy | if there was a *thing* to download called OpenStack that you could run on your fav distro and poke it with a stick, it would solve lots of this, but it breaks the main ecosystem we have built | 16:06 |
| notmyname | ttx: thinking of cultural differences in our community, I might translate your statement into 'murican: "john, that's the stupidest idea ever. plese never speak of it again!" ;-) | 16:07 |
| notmyname | ttx: I don't think it is, other than a cessation of focus on *how* the projects that make it up are doing things | 16:08 |
| ttx | notmyname: no, I'm actually genuinely confused | 16:08 |
| ttx | notmyname: the way i see it, the TC blesses a series of teams with a scope compatible with the openstack mission, and they produce deliverables, that are assembled in series on releases.o.o | 16:09 |
| ttx | and that is what "openstack" ends up being, a set of compatible building blocks for building open infrastructure ? | 16:10 |
| cdent | ttx that's conlfating intent with perception again | 16:10 |
| notmyname | ttx: that is true, but not complete (IMO). currently the TC also places a great deal of weight on the management of the code in the respective projects | 16:10 |
| notmyname | ttx: if it were only what you said, then the TC would have already packaged k8s and put it on releses.openstack.org as part of the things compatible with the openstack mission of providing cloud infrastructure software | 16:11 |
| ttx | notmyname: that's a fair point. In name of presenting a more... seamless experience of operating those components together, there are a number of things that are pushed horizontally | 16:11 |
| notmyname | ttx: for the record, I find that an admirable and innovative part of this whole openstack thing we've been doing eover the past 7 years :-) | 16:12 |
| ttx | notmyname: we currently primarily produce software building blocks, not really a distribution | 16:13 |
| fungi | the kubernetes team hasn't approached us _wanting_ to join openstack governance, have they? that seems like the key distinction to me | 16:13 |
| notmyname | ttx: based on the external voices you've been speaking with, what's the perception? is the perception that we are producing a distro? | 16:13 |
| ttx | although the Stackube project application ceratinly pushes the envelope in the distribution direction | 16:13 |
| ttx | notmyname: the perecption would be that vendros build distros on top of the framework | 16:14 |
| ttx | vendors* | 16:14 |
| ttx | I'll have to drop from the discussion as the day ends... but I loved this talk, let's continue over various media. | 16:15 |
| ttx | Found that the ML was a bit dead this morning so thought i would wake it up | 16:15 |
| notmyname | fungi: doesn't matter. python doesn't submit to RHEL governance to be included in that distro. I'm saying that if the TC defines openstack as just the stuff on releases.o.o, then it would have included k8s long ago. the point is that the TC *does* care about more than what's listed at that site | 16:15 |
| ttx | You are right that it indirectly asks the question of "what is openstack". But I don't think the current position is that much of a problem, if we fix the branding. | 16:16 |
| notmyname | ttx: in the interest of inclusion, I think it would be good for you (ie the tc chair) to link to this office hour's transcript for the ML thread | 16:16 |
| ttx | ack, will do now | 16:16 |
| notmyname | thanks | 16:16 |
| notmyname | what an invigorating discussion by which to start the day! :-) | 16:17 |
| * notmyname goes off to get ready for the rest of the day | 16:17 | |
| ttx | Opium does that to you | 16:19 |
| notmyname | lol | 16:20 |
| fungi | notmyname: okay, to rephrase, nobody interested in packaging and redistributing kubernetes has asked the tc to take them under our governance | 16:21 |
| notmyname | fungi: I don't see how the two are related | 16:21 |
| notmyname | or, wait. we're not disagreeing, I don't think | 16:22 |
| notmyname | obviously governance and distribution in openstack today are related | 16:22 |
| fungi | we can't govern nonexistent teams | 16:22 |
| fungi | nothing can be part of openstack from a governance perspective if there's nobody doing it | 16:22 |
| cdent | Isn't the underlying thrust here a reconsideration of what a team is? | 16:22 |
| cdent | and does? | 16:23 |
| fungi | something other than a group of people? | 16:23 |
| notmyname | my original comment is that the idea that openstack is only what is a tarball on release.o.o is an incomplete definition | 16:23 |
| notmyname | because the TC *does* care about governing the teams that have a tarball on that site | 16:23 |
| fungi | i still maintain that openstack is not software. but maybe that's the problem, we're conflating the name of a community with the name for what that community produces | 16:23 |
| notmyname | that is a very very good point | 16:24 |
| notmyname | openstack (the community) said "we've got a big tent" related to who. the external perception is "you've got a very messy big tent" related to what | 16:24 |
| ttx | notmyname: thought you'd appreciate: https://twitter.com/jaypipes/status/875377520224460800 | 16:25 |
| johnthetubaguy | +1 that hits on a very interesting issue vs the perception | 16:25 |
| fungi | the tc is governing teams of people, those people produce stuff, the tc is not governing stuff directly | 16:25 |
| notmyname | ttx: same song, 328th verse | 16:25 |
| cdent | notmyname: I demand you revise that statement, it is the 329th verse and unless you capitulate we will surely have a schism | 16:27 |
| * johnthetubaguy hands cdent and ice cream | 16:27 | |
| cdent | thanks | 16:27 |
| *** dims has joined #openstack-tc | 16:48 | |
| dims | duh, sorry i forgot to join this | 16:49 |
| cdent | you missed out on a party dims | 17:00 |
| dims | i know! | 17:01 |
| cdent | so, am I missing something, or do we not even agree on the definition of "big tent"? | 17:45 |
| gordc | cdent: it doesn't seem like it (i admittedly haven't followed big-tent stuff since conception and was confused back then as well) | 17:51 |
| sdague | well, that's the crux of the issue right? | 17:56 |
| sdague | big-tent is defined as all the projects under governance. For people steeped deeply in OpenStack, that means something. | 17:57 |
| notmyname | sdague: I think your proposal on the ML is a (relatively) simple and straightforward way forward without having to solve Once And For All, Again: "What is OpenStack?" | 17:57 |
| sdague | for the rest of the world big-tent means "everything I found at github.com/openstack" which isn't the same | 17:57 |
| sdague | notmyname: thanks! | 17:57 |
| notmyname | sdague: namely, mirror openstack projects to two separate github namespaces based on whatever bucket we the community put them in | 17:57 |
| notmyname | fixes the perception problem. is actually doable, and likely in a timeframe measured less than "years" | 17:58 |
| gordc | sdague: is that a messaging issue or something specific to the 'big tent' name? | 17:58 |
| sdague | notmyname: yeh, I'll admit, I don't know the technical limitations there with gerrit | 17:58 |
| gordc | or both given 2 years of confusinos | 17:58 |
| gordc | confusion* | 17:58 |
| sdague | gordc: at this point we're beyond a messaging problem, because this has become a fixed pointer in people's heads and any further messaging is bouncing off of that | 17:59 |
| gordc | sdague: understood. personally, i'm not sure who "here's a bunch of 'cloud' stuff the | 18:01 |
| gordc | proprietary companies provide but from companies that fund openstack | 18:01 |
| gordc | brand that is 'open'. we make no guarantees that they actually work | 18:01 |
| gordc | (they probably don't), this is solely to say these projects: exists, are | 18:01 |
| gordc | whoops bad paste. | 18:01 |
| gordc | i'm just not sure who this label is for ultimately. it seems ambiguous regardless. | 18:02 |
| fungi | it continues to sound to me like github is "the problem" here. i still don't know why we're promoting use of their proprietary service | 18:06 |
| fungi | i mean, i know it's because there are lots of people who think github is the only place they can find software and look for us on there, and if we didn't have a presence on it then someone else would be serving up copies of our software of dubious provenance and quality | 18:08 |
| cdent | fungi: getting rid of it is certainly an option (I'm trying to track the several that have come across the radar today) but I think it is medicine for the superficial issues, and doesn't address the deeper stuff | 18:08 |
| notmyname | fungi: how do you feel about http://github.com/mirrors? what about using that instead of our "by hand" ones? | 18:10 |
| fungi | i just wish there were a simpler way to redirect people to the canonical locations for our software and/or that github themselves catered to their users who want to "fork" git repositories hosted elsewhere on the internet so we wouldn't need to maintain copies there ourselves to satisfy them | 18:10 |
| notmyname | if I look at https://github.com/apache it looks different than openstack. ie the "mirror of <>" | 18:11 |
| fungi | notmyname: i hadn't seen that before, and it looks like a great answer potentially | 18:11 |
| fungi | notmyname: oh, wait, that's just a description of what we're already doing | 18:11 |
| notmyname | fungi: well it let's us actually push code to the most discoverable place (github) and potentially sidestep your philosophical concerns about it | 18:11 |
| fungi | notmyname: i'm not seeing what's different. http://github.com/mirrors just redirects to https://help.github.com/articles/about-github-mirrors/ which says (paraphrasing) "here are some popular projects which automatically mirror their source code to github" | 18:14 |
| notmyname | sure, in the DVCS sense there's nothing magic. I wonder, though, (and I can't find much info) if there's a different bit that twiddled somewhere in github so that github treats it differents | 18:16 |
| fungi | notmyname: or are you suggesting that we should switch from using gerrit's replication plugin to using per-repository post-receive hooks? | 18:16 |
| fungi | notmyname: oh, right looking at the mirrors they linked to from there i don't spot anything different from how our mirrored repos look on github | 18:16 |
| notmyname | it's mostly a discoverability thing. be findable on github, but also marked as a mirror to something else. and the way the kernel does github PRs is something I've heard discussed (do we do it?) https://github.com/torvalds/linux/pulls | 18:16 |
| notmyname | fungi: the "mirrored from <somewhere.else>" is the difference | 18:17 |
| fungi | notmyname: oh! now i see it | 18:17 |
| notmyname | compare the top of https://github.com/openstack/nova with https://github.com/apache/geode | 18:17 |
| cdent | even the icons are different | 18:18 |
| notmyname | anyway, if this is something interesting to our community, I think it's only part of an answer. I still like sdague's proposal of mirroring to different orgs in github for the different status of projects (and still having a common namespace in our community) | 18:18 |
| fungi | yeah, i was looking in the top-level list of repos in each org an dthe little "mirrored from" wasn't jumping out at me there | 18:19 |
| notmyname | hmmm https://stackoverflow.com/questions/11370239/creating-an-official-github-mirror | 18:19 |
| fungi | the icons might have been more of a tip-off if i dug down deeper than org-level | 18:19 |
| fungi | yeah, so far i haven't found any treatment of how you get that "mirrored from" status and repo type set | 18:20 |
| fungi | wondering if it involves reaching out to a human there | 18:21 |
| fungi | i want to say someone already asked them about that a couple years ago and they said they stopped adding them, but i'll try and find out more | 18:22 |
| sdague | yeh, would be good to know | 18:22 |
| sdague | it's also notable, that https://github.com/apache has less pages of projects than https://github.com/openstack | 18:22 |
| sdague | by 30% | 18:22 |
| sdague | and why digging in deeper would show that doesn't mean much, that top level view does matter to folks | 18:23 |
| fungi | yeah, earlier discussions with them turned up that you have to e-mail their support team for each repository you want set up, and that those are actually pull mirrors github has configured on behalf of those projects | 18:26 |
| cdent | the so link above confirms that | 18:26 |
| fungi | and not pushed via post-receive hook like the about page implies | 18:27 |
| cdent | (for at least the time of its writing) | 18:27 |
| cdent | and that they don't do that anymore | 18:27 |
| fungi | apparently you also needed to go to some effort to convince them it was in github's best interest to set up a mirror of your externally-hosted repository | 18:27 |
| *** lbragstad has quit IRC | 18:28 | |
| *** cdent has quit IRC | 18:32 | |
| fungi | so as some alternative solutions there we're brainstorming in #openstack-infra, we can renormalize all the repo descriptions to be something like "Mirrored from https://git.openstack.org/..." and add something similar (but probably more verbose) to the org-level description we've so far left blank | 18:34 |
| jeblair | notably, we can add a link to the *project navigator* from the org-level description | 18:36 |
| mordred | ++ | 18:36 |
| mordred | notmyname: (didn't see anyone answer you - but yes, we have a bot that responds to PRs people send) | 18:36 |
| notmyname | mordred: oh, cool | 18:36 |
| *** lbragstad has joined #openstack-tc | 18:36 | |
| jeblair | we could also somewhat tailer the repo descriptions (eg, "This is an $unofficial_terminology project..." vs "This is an $official_terminology project...") | 18:37 |
| mordred | notmyname: we've gotten some angry replies to that bot too | 18:37 |
| notmyname | mordred: "how dare you not accept my awesome code! don't you know that your project is broken without this pull request! I'm taking my code and going elsewhere!" | 18:37 |
| mordred | jeblair: yah. this is true. I think we'll need to rework the jeepyb gh integration a bit - but that wouldn't be a bad idea anyway | 18:38 |
| mordred | notmyname: yah. you've got it | 18:38 |
| jeblair | these things may all be worth doing, but i suspect whatever underlying problems we're trying to solve may extend beyond the set of folks who learn about openstack by browsing the repo list in github. (this may very well help with that set though) | 18:39 |
| jeblair | i confess, i do not know how large of a subset of "folks who are learning about openstack" that is. | 18:40 |
| jeblair | but these are good, not-too-difficult things to do regardless, i think. | 18:40 |
| mordred | jeblair: I agree - I believe the existence of openstack/ in the git link at all is the thing that is communicating information regardless of what verbiage might be posted to any of the web ui's - but I think we can post the verbiage and maybe it helps some percentage of the issue | 18:41 |
| *** gordc has left #openstack-tc | 18:57 | |
| *** hongbin has joined #openstack-tc | 19:04 | |
| fungi | i still like the idea of renaming that gh org to "openstack-and-pals" ;) | 19:07 |
| fungi | has a krofft bros. feel to it | 19:08 |
| smcginnis | fungi: https://giphy.com/gifs/movie-film-10Yay4O09aF5zW | 19:11 |
| fungi | ever | 19:12 |
| mordred | fungi: :) | 19:23 |
| mordred | I'll go ahead and go out on a limb and say that _probably_ we'd have people just as confused if we got rid of the github mirror AND the repo prefixes and only have git.openstack.org/nova - since openstack.org would be in the hostname | 19:23 |
| mordred | and "openstack the community in which people might host code" and "openstack the cloud software project" being distinct seems to be a concept that's alien to people | 19:24 |
| mordred | like- nobody was confused about openstack being on launchpad and thinking it was part of ubuntu | 19:25 |
| * smcginnis thinks they were already too confused by launchpad | 19:25 | |
| smcginnis | :) | 19:25 |
| mordred | heh | 19:25 |
| fungi | i wonder whether, instead of running two gerrits, we might be able to get away with two vhosts for git.o.o and something else and then put the governance-parsing logic on the git server end (maybe via a symlink farm)? | 19:25 |
| fungi | so continue mirroring to the git servers as usual, but make two different cgit sites and put their webroots in different directories with different sets of symlinks (or hardlinks even) built from governance metadata | 19:27 |
| fungi | maybe that's more of a discussion for #openstack-infra | 19:27 |
| mordred | fungi: well - I like it in as much as it's again a thing we could do without a ton of effort | 19:27 |
| mordred | fungi: HOWEVER | 19:27 |
| mordred | there is a compllication | 19:27 |
| mordred | fungi: (actualy, still might not be terrible) - and that's that in zuul v3 repos have a "canonical_hostname" that are associated with a connection | 19:28 |
| fungi | yeah, i was quietly hand-wavey about the ci side of things | 19:29 |
| mordred | so we tell zuul that repos found in the gerrit at review.o.o are "actually" from git.o.o- so that they get checked out at src/git.o.o/projectname | 19:29 |
| fungi | i think it gets easier with zuul v3 (people are probably long tired of hearing us say that) | 19:29 |
| mordred | so we'd need 2 gerrit connections each with a different canonical_hostname configured, and we'd need to move projects from one bucket to the other in the zuul tenant config if/when something moved | 19:30 |
| fungi | but still easier than moving projects between gerrits, or even repo renames in gerrit | 19:38 |
| fungi | i took a stab at jeblair's suggestion of updating the org profile at https://github.com/openstack to maybe clarify things some, but certainly if we tweak our manage-projects script switch the repo-specific descriptions out with mirror urls and feature the word "unofficial" in there based on a negative search against the governance projects.yaml that seems like it would help things further | 21:46 |
| *** sdague has quit IRC | 23:28 | |
| *** hyakuhei has joined #openstack-tc | 23:36 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!