OCI Weekly Discussion, 12 May 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: OCI Weekly Discussion - 2022-05-12

Description

https://hackmd.io/El8Dd2xrTlCaCG59ns5cwg?view#May-12-2022

A

A

uh It looks like it's me, so I can go ahead.

A

um So past two weeks we've been looking at cleaning up containers on the specs because we're making pretty good progress in the working group for reference types.

A

So um there was a pr that removed two of the seven maintainers on image specs. So now there are five.

A

And I had I proposed for sanjay and brandon to be put on there.

A

I think their work speaks for themselves, but people were then commenting on the thread.

A

What does it mean to like um be accepted as a maintainer who can nominate um rose had put a big comment on with good criteria? um Ultimately, I think, like the you know, a nomination from somebody within oci um should be pretty important, but maybe we can provide like as we're putting up these pr's for nominating.

B

A

Like kind of their accolades, so that people can look at the the history and be like oh yeah, that's why.

A

And I'm also unclear on to get a new maintainer if it's a quorum or what's the what's, the processor.

C

Is there a priority for um criteria like what folks should look at first.

A

I think no and that's kind of the question so um I'll try and pull up the comment, but I had I had basically uh proposed saj and then somebody had commented um not that sanjay's not gonna make a good fit but like what is?

A

How could I get involved- and I think there's not there's not a good guideline on like how people can get involved, and I think part of the problem is that we aren't getting things reviewed and merged, but um I think rose put together like a bunch of bullet points, and then there was another issue where akihiro had posted something about time zones where, like the calls, these calls are not doable in japan. So um maybe there's other criteria. We can look at. That's not just coming to these calls.

A

But I think just writing them down is what we should probably do.

C

Is a core contribution? A must.

A

um I think it's, it's really just an open discussion. I don't know I. I don't think there is a must right now. It's who do the maintainers want in, um I think, with image. Spec specifically, if you have a commit on there, it's very rare just because it has been has had inactivity. So I, like somebody pointed out said, like sanjay, didn't, have a commit, but that doesn't mean that he hasn't been looking at this stuff for a very very long time.

A

So um I don't know uh phil, are you? Are you here? I don't? Is there a good place to start with this? I don't know.

D

D

Well, first of all, I just add some um color, because I I think I think it was valentin uh rothberg who commented on sanjay's proposal- and I don't know if folks remember, but I just went back to try and figure out the time frame. But you know last spring. So, almost exactly a year ago we were kind of in well. We were in a worse quadrant that we had almost no like active image maintainers, and so we sort of did some uh well.

D

There was like a switch made, or actually I don't think that went through um there's like a proposal. There was a call and, and so three at least tycho valentin- and um there was a replace- I guess that's was jason, with john jason bazaar with john johnson.

D

So like we, we did this kind of big like hey. Let's, let's get this figured out and that kind of has helped us live the law limp along a little better and I think we're just kind of back to where that we knew that we didn't sort of clean up the whole process, because uh we still had a few folks that were totally um inactive, and so I think, your pr that's now merge fix that problem.

D

But but now we have the issue where, like three people a year ago, said hey I'll, be involved and at the end we kind of said no actually now that we kind of clean things up like we're good for now, and we close those prs and so now I think valentin is like well that's weird, because now there's two more proposals that are kind of like what we did last year um and I guess I'm wrong. Did he come in.

D

And, oh, I don't know where he commented that that was on you're moving.

A

Yeah, it's the crossed out one somewhere in there.

D

Oh god, yeah okay, but.

A

D

See it yep, I see it now, um so I think that's what generated the like? Oh well. I guess we need to have some kind of decision around. How do we decide who's.

D

Who should be up for maintainership or what are the requirements um so that you know, I think, that's kind of the backstory of? Why there's like oh wait?

D

We did this last year and we said no to a bunch of people and now maybe we're going to say yes to different people, um which you know, I think the I think what we're bumping into now is that just like any open source community there's not like a super easy way to say I mean unless, unless it's a a very fast-moving project, with tons of code contributions like specs just don't have like, I think your comment was was right on that. You know.

D

Sanjay and brandon had been here like very involved for a long time, but that doesn't translate into sort of usual things that we see in open source projects where we count commits or prs or issues or whatever.

D

So I I think, that's the sticking point here is like how do we come up with a criteria that makes sense for the kinds of activity we're doing right now?.

D

Josh, you think you have your hand.

A

Yeah, if, if someone else wants to jump in I the one thing I've been thinking with this is like it is really difficult to measure. um But I do I do think. The common thread in most cases is that other people in open containers uh feel strongly about the person. So there's kind of just like the soft like how you feel about it. Things, I'm wondering if we can put as just like a baseline, must have a nomination from a member of any any maintainer on any open containers repo, because then that at least.

A

That at least shows that somebody knows that you're doing something in some area. So, like someone, an image, spec may not know the things that people are doing in the working group, but that would be a valid. Someone from the working group could propose someone and then for people from the outside. They can try to.

A

You know buddy up with uh make friends with people on the inside to get that nomination. If they're interested.

D

Yeah I mean, I think, that's what raises an interesting point is like we have an interesting community and that we're like there's distribution, spec, there's image back.

D

There's the working group there's even tangential projects, there's people building registries around these specs and like the group of people that actually we you know you could take this meeting as a proxy for like the kinds of people who are interested in these topics, is a huge super set of like the five maintainers of image spec today, which is like at least what we're talking about today is image spec, um and so that that makes it tricky, because it's it's only those five people who have authority to approve or deny an addition or removal, um but this group of people that gathers and again there's a whole separate topic.

D

We probably should talk about that akihiro valentin, others feel kind of disconnected, because this is not a workable time for them for the most part. But so you know separating from that, like the this group that gathers and the people we see kind of actively involved in these discussions is not always related to the five inspect maintainers, and so that that's kind of an interesting disconnect that I think, makes us different than like.

D

I don't know container d or the moby project or some aspect of kubernetes, where the people that gather and that work on a single project are all kind of in a similar sphere.

D

Whereas here we have a bunch of constituencies, I think.

E

C

um So I have a suggestion about the time zone, difference um in other working groups that I've been in there's, usually a representative from one time zone like there are two meetings for the two different time zones and a representative from one time zone will be in the other time zone, and that representative is usually someone in leadership like someone who knows what's going on in um in one time zone and they basically give a rundown of what's happening in the other time zone, uh collect notes and then and then put that in meeting notes for the other time zone.

C

To pick up. Maybe we could duplicate this meeting for a different time zone.

C

I'll volunteer, if nobody wants to you, know, wake up early or stay up late.

D

D

Yeah I mean from an idea perspective like I. I think that that's great I mean I think that is we. We've had well, it's been several years ago now at some point. We had some period of time where we alternated, or we tried a different time, but if you do duplicate, I think you're correct that there has to be like somebody that draws the the line between the two. Otherwise you end up with two very disconnected groups, discussing potentially very different things so yeah I like that idea and I guess yeah.

D

I think it's up to like having someone willing to be that person to connect the dots.

C

Anyone else have thoughts.

B

I'll just say another: another solution to the time zone problem is to do more. Asynchronously and more I mean I think, we're pretty good about well as much as discussion happens. It can happen on slack or can happen on mailing lists, and that way we sort of reduce the importance of this meeting or or any synchronous.

B

You know, zoom meetings, that's another another approach, I don't know it's not necessarily one or the other, but uh more asynchronous communication on slack and issues.

C

Jason, in order to do that, someone has to take like we have to take copious notes.

B

And, or or or have this meeting less often or have fewer sl fewer synchronous, zoom meetings in general is another approach.

C

uh So what uh you think that most of the conversation could happen on slack for other folks to pick up.

B

I uh I don't know if I would advocate strongly for it, but it's another solution. I've seen work for the problem of of uh you know. This weekly meeting is not accessible to people in japan, for instance. If, instead of this weekly meeting, we had a monthly synchronous, zoom call and any other business happened on slack in email and issues that could be more amenable to uh other time zones.

C

So, like an all alternating time zone thing uh so one week we have it in ps in the americas, time zone amir and the other one. We have it in apac.

B

That works, if someone is is willing to, like you said both if someone in the apec time zone is willing to facilitate and lead that meeting, and if we have someone and either they they very meticulously, take notes and that other folks read or that we have a person whose job is to be that bridge or not job responsibility is to be that bridge.

B

I just think that's a that's a lot potentially to ask of folks. Unless someone is rushing to be that person, I wouldn't. I am unable to be that person personally.

D

Yeah sorry go ahead. Nation.

C

That might actually work where we have another meeting, but someone has in chaired by a person in the other time zone, but they have to you know, take down enough notes to be able to communicate with the folks in the in this time zone.

B

That would work if someone is willing to be that person. I don't know if I mean by the nature of this meeting. No one in this meeting is that person, so it's tough to sort of volunteer akihiro to be that person, while he's not here to defend himself.

D

I I guess the other twist on that is. Like um you know, alexa has been on the tmb calls last year when we had a few of them. We picked a time that worked with australia, um like I yeah jason you're right. We don't have them here, so we can't ask them to do this, but, like I could see alexa saying yeah like I'll lead a pac-time.

D

You know oci call, but I think one of the interesting twists is that, like alexa's focus is like run c like run c just had a release. A few days ago yesterday that alexa led, like his his mind, is like around run c and the runtime spec like.

D

If he let a call, I you know not that he can't cover other topics, but, like you know, his head is is more in that side of the oci, and so it's kind of interesting that this call has tended to follow kind of um where the bulk of the community has gone so initially like in five years ago. This call had a lot to to do with the runtime spec and then you know we get started into distribution.

D

A couple years ago and and a lot of the people who joined this call were were working on distribution, and so the other thing is like, maybe we're also kind of expecting.

D

I mean the title of this call is the oci weekly developers discussion or whatever we title it, and it's really been for about a year or two just about distribution topics and image, topics and artifacts, and so that's another. You know, and I'm sorry I'll answer this question in the chat since I'm talking, but you know that's that's up for debate like it was a suggestion.

D

It was a suggestion in the pr that start or the um sorry where's that link minimum criteria for new maintainers issue number 912., so reg judge commented in here I'll post it in the chat.

D

You know that that's potentially a minimum and I think what we're well, what akihiro said and what also valentin said in another forum, like that's kind of a tough requirement to put on people who, again because of time zones, could not participate here.

B

uh We need to do better about making that accessible to folks right, I would say we could just not make it a requirement right like unless, unless there is a groundswell of folks in other time zones that want to make another uh another time for this meeting, uh another solution to that problem is to just not make it part of the requirement.

D

Yeah and- and I guess you know my point- about kind of the the flow of this meetings- topics and discussion like if we were to have.

D

A sort of worldwide friendly, either rotation or once a month and have two versions of it like I think, there's a whole question. We're not. We can't even say that this is about maintainership in all oci projects, because you know run c, has their own set of maintainers, who obviously collaborate without ever coming to this call and they're part of the oci. So you know this is kind of a particular topic that we've been trying to solve about image, spec itself and even the requirements we put together.

D

You know by the oci charter that we don't the people joining. This call don't get to decide for the projects. It's the maintainers, so yeah nisha.

C

So it seems to me that the run uh runtime spec and the image spec are very closely um related.

C

So would it help to have a runtime spec maintainer like weigh in on image spec changes.

B

I don't I don't know if that I don't know if we can say that in the general case of all image, spec changes like, for example, the the type of image spec changes. We're talking about for reference types in particular have no bearing on runtime spec like runtime spec is inten is expected to. I think, as far as I know is expected to just completely ignore the new field, we would propose.

C

Okay, so then that's reason to believe that if.

C

C

If we make like maintainer decisions in image, spec.

C

Part of that is that the maintainer should be able to understand if it impacts either runtime or distribution.

B

Yeah I mean it might be worth having a runtime spec person, chime in to say officially on the record. You know with with the powers vested in them that this change makes no has no bearing on runtime spec just so that we can sort of check the box that someone from runtime spec looked at it and says that it doesn't matter.

B

There will also be changes in the future and presumably some type of change that is only related between image stack and runtime spec and the distribution spec is expected to just pass these things around on. You know blindly move this field from here to here. um In that case, it might be useful for someone from distribution spec to chime in on that change and say yep nope. We have nothing to to do with this.

C

F

So I'll put my runtime spec maintainer hat on, I would say that all the maintainers of the runtime spec have been involved in the oci long enough that, while changes to the image spec, don't necessarily affect the runtime spec, their opinion is probably at least useful.

F

It's been like I've been sad for months and months and months that there aren't more runtime spec maintainers on this call, but I think part of that is because the topics have been focused on image, spec and distribution spec for so long that it it doesn't necessarily feel worth some of their time to join. This call and spend time in this call saying nothing right when it's not like directly relevant, even though their opinion is interesting and useful and they've been around oci and have useful opinions around oci like it's.

F

I don't know it's hard, so I guess that's the long way of saying. Yes, I think their opinions are useful, but I don't think they're more useful than anyone else. Who's been in this call for months and months and months or been watching the github, repos or involved in slack or whatever.

C

So uh then tienen do you think that, with regards to maintainership, um like current participants would be able to make that decision without involvement from runtime spec maintainers or even like distribution, spec maintainers.

F

um Which decision.

C

uh Maintainers.

F

So like adding new image, spec maintainers- yes, I I don't know, I think, having maintainership on an oci project is- is essentially a way for the oci at large to say it. It has some degree of trust in this person in general, so I believe that, in my opinion, that holds some weight for like say if I proposed a new maintainer to the image spec, I think it holds more weight than someone who's never been to the meetings proposing someone as a maintainer to the image spec. If that makes sense,.

C

Yeah that makes sense to me.

F

But I wouldn't say that me proposing a new maintainer to the image. Spec holds as much weight as an existing image. Spec maintainer, saying hey. This person should also be an image spec maintainer, so there's definitely varying degrees. Here.

C

And so what happens when there are no image spec maintainers to make that call.

F

Then we make phil, take it to the tob.

D

Yeah I mean, I don't think we have that problem at the moment, because we now have five entries in the list that are at least around, and we know they're. You know they're paying attention for the most part.

F

Yeah, I think, for me at least the thing that's hard about. That is not what to do in that situation, but how to recognize that that's the situation we're in.

F

How do we know that the image spec maintainers are totally awol and not just this one? Github notification got lost in their inbox because that that's a real thing that happens to all of us. I would imagine, certainly me.

C

I um I have no response to that beyond you know repeated repeated comments that image spec maintainers are unavailable to review, prs or proposals.

F

Yeah from from my my view, as an outsider to the image spec but an insider to the oci, I thought that josh's pr recently to remove some maintainers while controversial on the face, was actually really good and really positive and managed to get some kind of response from all of the remaining image. Spec maintainers, if I counted correctly so that, like it, showed that there's some activity there and that I think the image spec is very much alive at this point today.

F

After that, pr.

F

I think it would be interesting to do a pr like that to the runtime spec that still lists michael at docker.com as a maintainer.

F

But maybe less urgent, because there's less important changes happening there, but there are pr's there that could use eyeballs from somebody who knows what they're looking at and knows what the implications of this will be. People like alexa, who probably doesn't have enough time to actually spend on reviewing runtime spec prs it's a hard problem realistically, so I I guess, I'm not proposing any solutions, I'm just throwing more problems in the pot.

F

C

Okay, well, um let's look at josh's. Well, it's an issue. It's not a pr! uh Last commented two days ago about a call and.

C

Besides akihiro, there isn't another comment from a maintainer and I am not sure if his assertion that asynchronous communications are working in this respect, because he's the only one that happened to notice.

C

Sorry josh go ahead. I.

A

I'm not oh, no, I just I just put it up. um I just want to surface uh something that I wrote in the chat here. Oh other things were being said. um I think having requirements is not.

A

I don't think hard requirements is maybe the answer here. So I think we can solve the time zone issue by saying: that's, not a requirement and the only requirement being that somebody from oci who's already has a maintainer position. Nominates you and then we can have based on rose's bullets.

A

We can have a template for these types of pr's and you just fill in the template and say this person does this and this and this number of years they've been doing this and this and the description of why you think it'd be a good maintainer, because I think 90 of this comes down to the other. Maintainers feel very.

G

A

Clicking the green button- it's not like, oh they made 90 commits, but I've never heard of this person it needs to be.

A

It needs to be um geared towards like uh triggering people's feelings. So I think I think we shouldn't have like hard requirements other than some baseline of needs come from maintainer and then yes or liking the template. So I think I think we can come to come up with a template based on that issue, and then we can close the the sanjay brandon prs and reopen reopen them with the template, and I'm I'm feeling pretty confident.

A

These two will be added, and my my big thing here is that the working group is where new stuff's happening and we don't have people on image. Spec who know about that. So having sanjay and brandon to be able to help us get things in was my primary thing, but to tianan's point yeah: let's we should do this clean up everywhere and just like keep moving things.

C

Okay, uh so I guess next steps is next step is to create a.

C

Maintainer, evaluation, template and a career like not korea, but like uh ladder for contributors to become maintainers or.

G

We don't need the ladder thingy.

A

Yeah, I think, specifically a pr template. I think, there's even in github, there's like a way to make that.

C

A

Can you can.

C

Yeah you can make that with the github templates.

A

Okay, awesome, yeah flavian. uh I hope I'm pronouncing that correctly, that they put a link. That's how to do that. So let's do that and I think I think we're good here and we can use that across oci too. Then.

C

Yeah I've done this before and if you want, I can submit a pr with a template.

A

Awesome fpr for the pr.

G

C

And actually, you know I've been thinking about collecting all these governance processes that we've been doing in the working group and over here and seeing if we can turn that into a into a standard for all three specs.

A

Yeah, that sounds great. We can use image spec as the guinea pig here and then um we can even go through tob if we want nisha and make that the common process for all the repos.

C

Yeah, uh let's see how it goes here.

D

Dude um so that all sounds reasonable is there?

D

Is there any kind of follow-up we want to do? I guess one of the things jason brought up. That kind of made me think you know there definitely are are pretty regularly gaps where we just don't meet, or only a few people connect here like.

D

Is it worth thinking about like a a more sporadic kind of regular meeting, whether it's twice a month or once a month, because then it might give us more of the chance to like have two of them like a apac, you know.

D

If we can find one, that's us plus europe, then all then you only need one other. You know apac friendly um anyway. I don't know if people want to keep mulling over that we don't have to make a decision. Obviously, but um you know it seems like the the call in recent times is used more to try and spur on activity just because there's been frustration about lack of things moving, and so the call at least gets people together who care enough to have the discussion and try and see how to make things make progress.

D

But if you have more active maintainers and things are moving in the sense of like async, you know pr's issues. Flow of conversation like this meeting you know, could go more in the direction of new ideas or people that just want to get together and you know do triage every once in a while. I think vinson had used it for that a few times anyway. Those are just some some ideas that maybe we can consider.

C

Maybe we can have a doodle poll to see who's available when.

D

Yeah yeah, that's.

C

Is that something we can mail out to the open containers, dev mailing list.

D

Yeah and I think.

D

I think amy has even done that on our behalf a few times you know from lf oci like management perspective, like of course that we don't have to have her do that, but um but yeah we can use the dev list.

D

To gauge like interest level for various time zones,.

C

Do we include the the relevant work to gauge interest.

D

You mean it in the sense of like what topics are being discussed here. Yeah.

C

What what uh seem to be the most common topics that come up.

D

Yeah or you know, is it worth saying here you know you can look at hack, md and kind of scroll and see kind of the flow of content, but you know maybe, additionally, for polling on time zones pulling on.

D

You know: are there issues oci areas of interest that never come up in this meeting and if those topics were here you know five more people would would come. I don't know that gets a bit trickier, because doodle doesn't exactly support that kind of uh polling, but.

C

Also, no one takes notes in this meeting.

D

Yeah you tried to help us nisha and we've already failed. You.

C

Well, there's no, uh so the reason why it works in the working groups uh in the reference type working group is because there's a governance document that says somebody will be a note taker and we have pretty gold and silver medals.

D

C

And we and we say, like we, really love you no take us. Thank you so much.

D

Yeah, no, that's that's good good point. We haven't culturally made that a um a worthy cause here, obviously.

C

D

C

The whole point of a governance document is that folks need some kind of direction on how to how to run these meetings, if, like folks, just join and sit and wait.

D

Yeah yeah plumbing.

D

That's a good point like you know. I think you had done this for a meeting or two like. If, if step one of any meeting is like who wants to take notes, then you know you kind of build it into the structure of the meeting.

C

Yeah, that's uh that's how the working group also does it, but um I think we've had it enough times that the first thing that happens is somebody says I'll. Take notes.

D

D

D

I guess we did it. I was just looking at. I guess we do have it in the template now, but we're not. I think the missing piece is kind of uh incorporating that into the meeting flow. Just getting that done. First thing.

C

I'll be happy to be the person who brings that up. The very first thing.

D

Yeah, no great that would be good. Okay,.

C

I'll do that for a little while and let's see how it goes from there cool.

D

um Any anything else folks want to discuss, so we give folks a few minutes back.

A

um I just want to say thanks everybody for trying to make progress on the maintainers thing. It seems small that, um like I think, there's a lot of people who want to do things in this community, but they can't so. I think we can keep making good steps in that direction.

D

Yeah yeah! No thanks to folks here I know: josh you've been fighting that fight for a good while um so yeah thanks. uh It's hasn't always been the easiest community to move forward, but yeah. It does seem like we're.

D

Piecewise getting things in place to to make things smoother. So that's good thanks to everybody.

D

All right cool! Well, um I guess for those uh I don't know if anyone here is headed to kupcom, but I know vincent and I should be there.

D

Feel free to say hi if you're there and it may mean uh well, it may not mean it may not mean much impact to this group next thursday. But I know I won't be here and vincent can't so, but catching a couple weeks personally.

D

C

D

Thanks, everybody have a good day.

D