2024-03-21 Meeting notes

 Date

Mar 21, 2024

 Participants

  • @Feanil Patel

  • @Kyle McCormick

  • @Jeremy Ristau

  • @Chintan Joshi

  • @Robert Raposa

  • @Tim Krones

  • @Adolfo Brandes

  • @Michelle Philbrick

  • @Piotr Surowiec

  • @Felipe Montoya

Previous TODOs

 edx-platform Discussion topics

Item

Presenter

Notes

Item

Presenter

Notes

Start Recording

 

edx-platform assets

@Kyle McCormick

edx-platform write access

@Kyle McCormick @Jeremy Ristau

  • Teams in CODEOWNERS require write access. That means that if we want to use CODEOWNERS to list edx-platform app maintainers, we would need to grant write access to every listed maintainer.

    • Possible workarounds

      • Instead of listing a team, list the leader/manager of that team

      • Instead of listing a team, list an edx-platform CC on the team. This, obviously, requires that at least one member of the team be a CC on edx-platform, but maybe that’s OK?

      • Just give the repo write access to the whole team. How would that work with the CC program?

      • Require reviews from CODEOWNERS. This would limit write access of everyeone in the repo to the CODEOWNERS list.

        • We would need to grant write access to edx-platform CCs in codeowners as well.

      • Avoid this issue--don’t use CODEOWNERS to manage app maintainers. Use something else.

  • We don’t need to figure this out now. Revisit once Redwood upgrades cool down.

 General Discussion topics

Item

Presenter

Notes

Item

Presenter

Notes

Start Recording

 

edx-platform Python Upgrade

 

@Feanil Patel

  • Python 3.11 instead of 3.12

  • Numpy and SciPy Upgrade

    • No Numpy that supports both Python 3.8 and Python 3.12

      • Means that we need to target 3.11 instead of 3.12 for edx-platform

    • Both Numpy and scipy are in edx-sanadbox (codejail env)

      • JB: We’ve had separate requirements files to smooth transitions

        • Feanil: Yes

        • We could upgrade edx-sandbox to 3.11 for Redwood, but 2U and others could fall back to the 3.8 sandbox for a while if they need to

        •  

    • Will impact 2U earlier as it lands on master.

Still need folks to pick

up python 3.11/12 upgrade work

 

Should 2U PRs go to the Contributions Board?

@Michelle Philbrick

  • For repos where they are not the maintainers, that’s a fine way to get them reviewed.

  • This came up when Adolfo pointed out that a certain PR authored by a 2U employee needed some product review. Question became “what’s the process here?” How do they get that product review?

  • If we tracked 2U contributions on the board as well as everyone else’s, that would add a lot of weight to the contributions board process.

  • Feanil: Ideally we’d have more people helping with the contributions board.

  • For the time being, leave it up to 2U owners to get product review.

  • Michelle: The change should already be on Jenna’s product review board, so we shouldn’t need to add it to the contributions board.

  • Feanil: Still, there will be 2U PRs that are blocked by code review on repos they don’t have write access to. In other words, there will be more OSPRs, and our process will need to adapt

  • Robert: If we have a single 2U PR that is blocked by community code review, could that single PR get onto the contributions board, without ALL 2U PRs going there now.

  • Michelle: Tim is maxed out w.r.t. time spent managing the contributions board, Michelle nearly is as well, the process can’t handle more PRs right now

  • Tim: Back in the maintainership pilot days, we’d said that the contributions board would eventually go away (or at least not need manual management), because maintainers would be able to manage incoming PRs.

  • Michelle: Would that ^ process have a way for people to look in order to figure out who is going to review their PR?

  • Adolfo: Maintainers should be running this process

  • Tim: We need some way to automatically tell people who their reviewer should be. Some sort of automation, looking at catalog-info.yml, or linking to backstage

  • Feanil: We should start scaling down the contributions board, perhaps by removing repos that are maintained?

  • Feanil: We should recruit a 3rd person to help Tim and Michelle

 Action items

@Feanil Patel Figure out edx-sandbox (codejail) py38 vs py311 technical details
@Feanil Patel Figure out whether we need to test 3.11, 3.12, or both for various packages.
@Michelle Philbrick try to recruit a 3rd contributions triage person
@Michelle Philbrick @Tim Krones @Adolfo Brandes Work on two docs: Maintainers expectations and training, and message for contributors to orient them to our process and expectations. Existing How To Maintain doc.

 Decisions

  1. edx-platform Redwood will be targeting 3.11, not 3.12
  2. We will need to test edx-platform packages with 3.11
  3. We need to continue compiling
  1. We will have multiple edx-sandbox (aka codejail) requirements files in edx-platform: py38.txt and py311.txt. This will allow operators to use 3.8 in the codejail sandbox temporarily after Redwood. More to figure out technically there.
  1. The Contributions board is not the long-term strategy for triaging PRs. Maintainers should be doing that for their own repo.

 

Recording and Transcript

Recording: https://drive.google.com/file/d/19YXvuwPbBTP4LQLuKr5hn3FCURcWrpVE/view?usp=sharing

cbx-ybic-gyq (2024-03-21 09:32 GMT-4) - Transcript

Attendees

Adolfo Brandes, Awais Qureshi, Chintan Joshi, Feanil Patel, Feanil Patel's Presentation, Felipe Montoya, Jeremy Bowman, Jeremy Ristau, Kyle McCormick, Maria Grimaldi, Michelle Philbrick, Piotr Surowiec, Rabeeh T A, Robert Raposa, Tim Krones, Yagnesh nayi

Transcript

This editable transcript was computer generated and might contain errors. People can also change the text after it was created.

Feanil Patel: All right morning everybody. Yeah, that's good going. There's I don't think too much to talk about today. So let's start to Do's from previously. On the top is mine, which is the dock for models for maintaining its platform. I haven't gotten around to that yet. But it's getting closer to the top of my list this week. So I will hopefully have something to have people to start putting ideas into later today or tomorrow.

Feanil Patel: Robert and I guess letting slash Adolfo. Do you guys have a good handle on the MFE deployment plan for the new 20 upgrade is that?

Robert Raposa: I mean I can't.

Adolfo Brandes: Yeah, we have I mean nothing's happened this past week about it and…

Feanil Patel: right

Adolfo Brandes: it's not likely to happen next week either but we have an agreement on what?

Feanil Patel: Okay, cool. but

Robert Raposa: Yeah, I'm gonna check my side of that off. Does that make sense it up?

Feanil Patel: yeah.

Adolfo Brandes: Yeah, yeah. Sure.

Feanil Patel: Yeah, if you guys have a communication plan, I don't think we need to keep checking on that as long as you guys have a know what to do when those things happen.

Adolfo Brandes: It's just that on the front-end sides. The plugin stuff is actually taking priority right now. But once that's over and…

Feanil Patel: get you

Adolfo Brandes: done with which might mean once read with this over and done with that's top of the list again.

Feanil Patel: yeah.

Feanil Patel: That sounds good. Yeah, I mean, I think We're not bumping up against the security problem with no 20. So I think upgrading as needed to know one day seems like a fine. plan

Adolfo Brandes: Yeah, the main mode of immediate motivation now is running into packages that don't work with no dating anymore like semantic release.

Feanil Patel: Right, right.

Adolfo Brandes: And we haven't been cornered yet by that, but I am sure we will reasonably soon.

Feanil Patel: Yeah, I think the plan should be to just upgrade. when we need to but

Feanil Patel: I think we'll need a plan to be able to deploy some of these things on 20 with other things being deployed on 28.

Adolfo Brandes: For sure, this will require some coordination at the very least with the libraries right front and…

Feanil Patel: Yeah.

Adolfo Brandes: platform front and…

Feanil Patel: then red

Adolfo Brandes: build and everything.

Adolfo Brandes: So what I can say that might make this easier is that I enabled the alpha builds or all the basic Upstream by components and libraries basically front-end build platform components header and footer. So that we can start using the alpha Branch to make these kinds of breaking changes easier because then you can publish a package and then actually use the alpha package. Downstream to see…

Feanil Patel: I see. That I got it cool.

Adolfo Brandes: if it works, right?

Feanil Patel: Is that documented somewhere?

Adolfo Brandes: Not right now. I just ran with the idea. We were actually discussing. the front end build Alpha package, which already existed for typescript …

Feanil Patel: Yeah.

Adolfo Brandes: and we wanted the alpha test just 29. And it says okay, let's do both together. And that's what we're trying right now. And that prompt is like enabling the other Alphas.

Feanil Patel: That's got it.

Feanil Patel: Okay, that sounds worth.

Adolfo Brandes: So it's coming up organically,…

Adolfo Brandes: but it'll be coming in ADR.

Feanil Patel: Yeah. Yeah. I meant more how to use Alphas in your Upstream MFE kind of more than a decision.

Adolfo Brandes: Yeah, exactly. Yeah. we're sort of yeah.

Feanil Patel: it's not hard but

Adolfo Brandes: Yeah, but we're establishing some patterns right now.

Feanil Patel: Yeah, exactly.

Adolfo Brandes: Just so that the alphas are predictable and we know what the contract is or is not in particular,…

Feanil Patel: Right. Yeah,…

Adolfo Brandes: right?

Feanil Patel: All right to Jeremy ristau, you have a couple there do is that still helpful to keep those on here, which is timeline for Aura MFE.

Jeremy Ristau: So I think that one is as done as it needs to be for this to-do list. I posted a Confluence page in the maintenance working group slack Channel.

Feanil Patel: Okay.

Feanil Patel: Yeah.

Jeremy Ristau: so any additional needs are The follow-ups to just go in that Confluence page.

Feanil Patel: That sounds good. Yeah, and I linked Jenna to that Confluence page. So hopefully the product working group will be on there for any questions they have

Jeremy Ristau: .

00:05:00

Feanil Patel: I'll check that off the internet

Jeremy Ristau: the next one related to the handoff of certain parts of the upgrade process. personally. I think that's gonna be much more of an organic thing.

Feanil Patel: Yeah, I think it's right.

Jeremy Ristau: But in terms of an audit of all the steps that's definitely not at the top of my current set of personal to Do's so

Feanil Patel: I kind of feel like this thing will just happen as problems arise rather than as push from you at this point and…

Jeremy Ristau: yeah.

Feanil Patel: I'm okay with marketing it off given that

Jeremy Ristau: Sounds good to me.

Feanil Patel: and then ownership audit.

Jeremy Ristau: so that one is in progress. I would say there's three main buckets for it. There's the sort of learning platform section. There's the marketplace technology section and then there's the Enterprise section. I'm moving out of the learning platform space and into the other two, so, it's a bit of a trickle because it's just me doing the follow-up…

Feanil Patel: but Brad

Jeremy Ristau: but it's certainly in progress though.

Feanil Patel: Yeah, awesome. I'll leave that one in place then for now.

Jeremy Ristau: but I will say this effort did generate a Follow-up point that we talked about earlier today related to code owners for edx platform. So that's the second or…

Feanil Patel: right

Jeremy Ristau: the third row of this table if people are interested in reading the notes related to that.

Feanil Patel: Okay, cool Yeah.

Robert Raposa: And one thing of it we can check into this later, but the automation that you have to pull the catalog information into a spreadsheet.

Feanil Patel: Yeah.

Robert Raposa: Is that something shareable?

Feanil Patel: 

Feanil Patel: Yeah, it's literally like a headline script. I wrote I copy-paste the column of the repos into a text file and then it gives me a text file with the corresponding ownership catalog info. So I'm happy to send that around but it's not super complicated.

Robert Raposa: right Yeah, yeah.

Feanil Patel: Yeah, yeah, I'll post that after this meeting.

Robert Raposa: Okay. thank you.

Feanil Patel: Yeah.

Feanil Patel: all right, and then Adolfo, do you still need this ticket on here for enabling the note 20 testing on mfe's? He said it's back Bernard. So maybe it doesn't make sense for us to be drinking it on.

Adolfo Brandes: About back burner and we actually have tickets for the main mfis already.

Feanil Patel: It okay.

Adolfo Brandes: It's just a matter of finding a good time to execute.

Feanil Patel: Okay.

Feanil Patel: Yeah, yeah. Okay. That sounds like It's not super valuable to have this checklist item here at this point that specially great.

Adolfo Brandes: No, no, no.

Feanil Patel: So I'm gonna check that off and we'll all that done for now. Then we went to extra wide mode.

Feanil Patel: That's right. didn't think there's only one major thing. I wanted to check Canada today, which is I've been spending some time on the nude or nothing on the Python 3 12 upgrade for edx platform and I discovered that the

Feanil Patel: the Sci-Fi the numpy package. There is no package version that supports both python 38 and python 312 which is bad news for us because it means that we have to hit Target 311 for NX platform instead of 312.

Feanil Patel: Because there's no simple way to make that job the other corollary to that. I think this specifically impacts. to you sooner than Redwood is the numpy and sci-fi packages both have to get upgraded and They are both in the code jail sandbox. So whenever we do update them, there are potentials for breaking.

Feanil Patel: python code generated problems in the platform so I would love to sort of talk about that Jeremy B. What's up?

Jeremy Bowman: Yeah, I just want to note that have in the past for Django upgrades actually had either second requirements files or requirements with conditions attached to them so you can install this numpy version with 3/8 and this numpy version with 312. So the same code base could be used you just be installing a different version of numpy depending on which version of python you have.

Feanil Patel: Yeah, so I think that's one of the strategies I was thinking of using which is that we can leave there's a three eight dot text file that's there right now for the sandboxes and we can leave that in place and maybe create a 312 file that would have the updated version what that would allow is for anybody. For that upgrade to sort of hand an independently and for the community we would default to 312. sort of by Redwood, just in general and then The deployment could make that jump whenever you guys wanted to with it, but probably before sumac I would say and then we'd get rid of the three eight. At sumac it's a little risky because the version of python running inside of that code jail would be the older version which wouldn't get security updates. So that's like a risk that anx.org would be taking on for as long as you're running the old sandbox.

00:10:00

Feanil Patel: but that level of flexibility I imagine is useful so

Feanil Patel: that won't be too hard to do but that's kind of the big thing. That's like another piece that Everybody will need to deal with is when they upgrade it as platform their code JL sandboxes will get a newer version of numpy sci-fi and a bunch of other packages.

Robert Raposa: And if I understood correctly, there is a 311 possibility that would not have the security problem. but we still need to get off of whatever we're on and get on to what the overall. that

Feanil Patel: yeah, yeah, so my proposal with this is that actually we target 311 for edx platform just so that we can because otherwise we would have to make a breaking change. All at once which isn't hard for. It's tricky because testing on both versions it gets a lot more complex. If we need to have one python for numpy version for one and the other for the other if we target 311, we can achieve a set of dependencies today that work that theoretically work with both.

Kyle McCormick: I'm just going to say that tooth The Redwood python 311 for both edx platform main environment and the edx sandbox kojo environment. With a requirements file that works with python 3.8.

Feanil Patel: Correct.

Kyle McCormick: So people can run that.

Feanil Patel: For the sandbox if they want to yeah. Yeah, yeah.

Kyle McCormick: cool

Feanil Patel: A corollary of that is that for a bunch of our sort of shared Django utility related libraries. Those will need to support 311 and 312 if they're being used in edx platform and in other Django services. And for any Library that's only used in metix platform rather than targeting three 12 like we've been saying so far. We would be targeting 311. So we need to update some tickets and do some communication around that.

Kyle McCormick: If the library works with 312 will it work with 311 or…

Feanil Patel: this

Kyle McCormick: other breaking?

Feanil Patel: I think we have to test both. It should but I don't think. But the other lessons that I learned as I've been doing this is that we want to make sure that we're still compiling the requirements with three eight and that they install in 312 instead of compiling them for three 12 and hoping that they still install them three eight. So essentially holding things back so that they'll continue to work with three eight eleven and twelve

Feanil Patel: while we're doing all of our testing and then after we've sort of gotten three eleven working everywhere and it becomes official then we go back and clean up we update all the requirement upgrade jobs to run with three 12 instead of three.

Kyle McCormick: Cool, and one last question on the sandbox thing.

Feanil Patel: Yeah, I don't want to introduce it right now because I think it'll complicate this. It should be pretty easy to add a couple of lines to the edx platform make file right now to get the 311 or 3/8 requirements generated. I think what ends up happening and correct me if I'm wrong Jeremy, but essentially a Jeremy Bowman essentially. we would stop upgrading the 38.text whenever we start using the 311.text.

Jeremy Bowman: Are you thinking about just for the main environment or from a code jail perspective?

Feanil Patel: I've just for the code jail environment for the main environment. I want to be able to compile it 3/8 but have it be installable in For the code jail environment. I want to make sure that there's a requirements file that doesn't upgrade sci-fi numpy and any other major packages. I know version that does. So I was thinking that we would have to stop running makeup grade on the 381, but I guess we could just add more constraints to the 381.

00:15:00

Jeremy Bowman: yeah, like you if you're willing to have two different make upgrade jobs. You could still update the 381 if you wanted to.

Feanil Patel: All right. I think there's a little bit more technically to figure out there. But I'll spend a little time on it this next week and…

Jeremy Bowman: .

Feanil Patel: see if I can come up with something.

Feanil Patel: That said there's still a lot of libraries that need to be upgraded and support 311 and 312. Thank you to chintan and a couple of other people who have been trying to make some of those move forward even though we're learning a lot.

Feanil Patel: But if other people have cap cc's of capacity, that would be really useful is to pick up. The python upgrades for various libraries and be opened X get up work. And if you don't know which one's I will link them.

Feanil Patel: You'll find it.

Feanil Patel: They're all linked from in.

Feanil Patel: I'm hoping to spend a little bit more time on those tickets and groom them up so that it's actually a lot more accurate whether we need three 11 through 12 or both for them. So I'm gonna try to do that later today, which is for any libraries that touch edx platform and another service. We'll need the testing to run 11 and 12.

Feanil Patel: And somebody else has capacity to help do that. I would love your help with it. Otherwise, I'm going to be doing it.

Feanil Patel: right other questions

Feanil Patel: Michelle

Michelle Philbrick: Yeah, I'm not sure if this is the correct venue to bring it up. But Kelly Buchanan asked yesterday if you should start adding PRS to the contributions board and I wasn't sure about the contexts for it or the reasoning behind it. But I wanted to bring it up in case it's something that's been kicked around because of maybe the new maintenance items and things like that, so I just wanted to See if that's something that's been discussed or if it's something that we need to talk about further because I wasn't sure about the background of it.

Adolfo Brandes: I have a little context on that.

Michelle Philbrick: Okay, cool.

Adolfo Brandes: This is in relation to a PR that I ran across that made a small user-facing change. and as I've been doing. I pointed out that that might need product review.

Adolfo Brandes: But it's a small Improvement. It's not really controversial, but it raised the question. What do we do about this kind of thing? what's the process here? Right. Also, I figure I didn't talk to her specifically about this. Afterwards, but I figured this is a way of her asking the question …

Feanil Patel: right

Adolfo Brandes: what is the process and I don't know myself like I asked product. For these small PRS what do you want us to do? Right and the answer was we have a status. For this kind of small Improvement, but there's already a PR in the product board or in the roadmap. But I don't know what pace these PRS are gonna be reviewed and what happens if they get stuck. I don't know So maybe Kelly is trying to find a way. To navigate this right?

Feanil Patel: Yeah.

Feanil Patel: Yeah, and relatedly the point that I thought you were making Michelle was that since to you is not maintainers on more and more boards. They will need other people in the community to be reviewing work and in that. Space how do they get those reviews it probably starts making more sense for those PRS to get on the contributions board where they are making changes to things they are not the maintainers of also so it sounds like there's a couple of good reasons. Why To UPR should get on the contributions board.

00:20:00

Feanil Patel: And we should update the bot to perhaps. Just put all. non-bop PRS on the contributions board

Michelle Philbrick: Okay, I think Tim I don't know if you want to jump in but right now we're still only triaging open source. PRS with that still be the case,…

Feanil Patel: right

Michelle Philbrick: even though to use would be added just because that would be a lot. Yeah.

Feanil Patel: So I mean, that's a flip side is that It I think the capacity that we have right now and we might need more capacity for those to get review they would get reviewed slower. So. I don't have an obvious answer today. I think in an Ideal World, we would find one or two more people to help you guys triage things, but then all of the two UPR so go through that same board and go through the same process as all of the rest of our PRS. I think that'll make it easier to do things. Like Adolfo is saying which is confirm that they've gone through the community product review or have been accepted as a thing that doesn't need a product roadmap ticket without sort of unifying this process. some of this becomes harder to achieve successfully

Michelle Philbrick: And then for the meantime since it is just him and I could we leave it that it's up to you owner to get reviews.

Feanil Patel: Yeah. Yeah,…

Michelle Philbrick: the moment

Feanil Patel: I think that's fine. I think perhaps we can communicate inside to you Robert and Jeremy that if you do need products help and you need to navigate the process. You can help maybe you can tag Michelle and Tim on that pull request so that I can get integrated into the process. And then Michelle, maybe you and Tim could think about if there are or just do it post to try to recruit more people to help you guys triage. Is that a crazy idea?

Michelle Philbrick: So I have a quick follow-up question. Is this just for PR's that may need product review or is this for all two uprs?

Feanil Patel: I think yeah. Yeah.

Adolfo Brandes: I would stick to product. once at least for now

Michelle Philbrick: so in that case and Tim's please jump in if you have other suggestions, I don't necessarily know that they would need to be on the contributions board because they're already going to be on Jenna's project board and the new product process actually has product review happening before those PRS actually end up on the contributions board unless we catch ones that have fallen through the cracks. So I'm not sure it makes sense Add If it needs product review because it should already in theory be with products that make sense.

Feanil Patel: So yeah, if it means product reviews saying there would be a ticket on the product board or we would need to be making a ticket on the product board before it even gets to the contributions board as you guys are saying. Yeah.

Michelle Philbrick: Yes it shouldn't be on there until product has reviewed it anyway. Yeah. according to their new process obviously Adolfo,…

Adolfo Brandes: Okay, I might sense, right?

Feanil Patel: Got it. Okay.

Feanil Patel: right

Michelle Philbrick: you've done an awesome job of kind of going through and weeding out ones that we didn't catch sooner and I'm sure that will still happen. But for now, I' if it's related to product review. I'm not sure they need to end up on the contributions board. before they need product review

Feanil Patel: Yeah.

Feanil Patel: I'm

Adolfo Brandes: Sounds like we have an answer for or

Feanil Patel: Yeah to me this brings up another point though, which is that eventually. Even if they've had product review if the contribution is in a repo that to you is not maintaining. Then those two UPR still need to go through essentially the community review process.

Michelle Philbrick: Yes, yeah.

Adolfo Brandes: Let me just raise a small like.

Feanil Patel: And that will be more.

Adolfo Brandes: Even if to you are maintaining the report they still need to go through the product review process for contributions in there.

Feanil Patel: Yeah. No, I wasn't saying I was saying they like the maintainer review process. If to use the maintainer,…

00:25:00

Adolfo Brandes: Okay.

Michelle Philbrick: Just in general.

Feanil Patel: they can do the maintainer review process.

Kyle McCormick: sorry, my other point of information we maintainers don't have a blocking review on their So any repository that they don't have

Feanil Patel: Exactly. So as to you is reducing the number of things they are no longer the maintainers of all and won't have right access to a bunch of repos, but they have today right and they might still want to make changes in those repositories and make pull requests that the maintainers of those repositories. Will need to review and rge. In that case, we will want those to be tracked in the same way as we track other open source contributions from other companies. And that is essentially me saying that we're about to get more osprs that will need to be processed and managed. Then we have had historically because of the ownership shift. and so Michelle and Tim this is sort of like a warning Bell for you guys that

Feanil Patel: Given that this how should we change the process or how should we recruit do we need to recruit more people to sort of coordinate? Do we need to change how we're doing those processes? We don't have to have an answer today, but that's stuff that we should figure out Robert I think first.

Robert Raposa: Yes, so in the short term. I think if we assume the simplest question of hey, I'm to you I have a PR that's of this nature and it's blocked and I'm waiting on a review and I simply want to know what is the process right now for me to unblock it by getting it onto the contributions board that feels like a very simple problem. That's not going to overwhelm everyone because to use Going to know that they need to do this process and it's not like you have to add every single PR there. So if we start there, I think the question would be What is that process how do we get that over to the board and Michelle so that We're aware that that needs to move through.

Feanil Patel: Threat and I think the sort of corollary I would add to that is we need that process for today. And also assuming that you will have more pull requests in the future and the number of these might increase what if anything do we want to change about the existing process to improve it to be able to handle that in the future? So I think those two questions are aware of Michelle and then Tim

Michelle Philbrick: so I think Tim is already maxed out as it is and…

Feanil Patel: are

Michelle Philbrick: he can confirm that with his contribution hours. And for me, I am just worried that this will become my day job, but because right now I already have hundreds and hundreds and hundreds of pull requests that I go through and I think in the meantime if these are being added to the board and need riage it's going to take a lot longer for things for us to comb through them and everything. So I would just keep that in mind Tim. Do you want to jump in?

Feanil Patel: before Tim

Tim Krones: Yeah, So I think I agree with what Michelle said. I'm thinking back to what we said initially when we came up with Holcomb's content of ospr triage. And when the maintainership pilot was still going on and we were saying once everything is in place and working as it should be then there will be no need for Michelle and I to be doing what we are doing now, which means we need to documents the processes and empower.

Tim Krones: Pull request authors to do as much of the process themselves as possible right where it's just Michelle and I just jumping in to resolve cases where there's questions or where there's somebody who happens to not know what the process is, but it won't scale if it's always the two of us or even if we add a couple more people I don't necessarily think that's the right solution in the long term.

Feanil Patel: Yeah, so that sounds like a great process and it sounds like maybe we should start investing more time in that now. And figure out what the next steps are there, Michelle.

Michelle Philbrick: The only a comment I have about that is. I guess it's maybe just a clarification for Tim but

00:30:00

Michelle Philbrick: Sorry my brain for that process would that include somewhere for people to look so they know who's Vapor pull requests because right now it's us directing the pull request to the proper people for review. So I'm just curious how that would work. But again, that could be a conversation for another day if we're gonna work on that.

Feanil Patel: I don't know.

Adolfo Brandes: My thoughts on this is that the job you're doing Tim and Michelle should be at least in my view ultimately the maintainers responsibilities. So whenever a PR comes into their repo, they should, get notified. They should have set up notifications to get notified when a PR comes in and they should know the process and then they should direct what happens with that in other words triage, right?

Feanil Patel: Tim

Tim Krones: Yeah, I agree with Adolfo said as well. I think if we can approach it from both ends that will be great educate the maintainers and also do a better job of educating the community members that are sending in contributions. There's another level to it. Maybe this is more to Michelle's point is how do we surface the current information about who is the maintainer? Who should I paying if I'm a PR author? How do we surface that on the pr as themselves? And I think we talked about this a little while back in some thread. I don't know but it should be possible to automate that a little bit right? We have the catalog info files. Somebody raises a pull request. It should be possible for a bot to say if you want this reviewed paying these people and first you have to have a green build Etc, put in some of the other rules that

Tim Krones: that Michelle and I sort of Applied manually when we do triage.

Feanil Patel: Yeah, Kyle.

Kyle McCormick: yeah agree with everything that's been said so far I think we say a lot of maintainers should do this. And this should do that and it's true. I think we'll still need someone or some group kind of keeping an eye on. has maintainers it's hard burned out get distracted have someone saying hey, there's a backlog up here as piling up. Do you need help? Do you need to be swapped out as maintainer? And so we will need kind of some sort of process looking at that and doing that but yeah long-term is really agree. we shouldn't need this board that has two folks spending a lot of their time manually managing it and manually taking care of these PRS.

Feanil Patel: Yeah, so I think that's a great Point Kyle I think.

Feanil Patel: It's a good reminder that this was meant to be temporary and I was sort of solving in the space. We already had rather than where we were intending to go. So thank you for that given that we eventually want this to be a maintainers responsibility and I think as you said I was like, right. It's actually written down in the maintainers responsibilities that they need to be responsive and these things I wonder if the work we should be doing on the contributions board is to be shrinking it to the set of systems that aren't maintained as we find maintainers for more of the repos. and also building out the training for maintainers that they can take to be prepared and understand what that day to day in the life of triaging and maintaining the repo looks like is not everybody has done that before.

Feanil Patel: And Tim and Michelle maybe I could ask you to sort of think about and draft up something that would look like that training just an outline to Go forward Tim.

Tim Krones: Yeah, I think I would be up for that one conflict and Michelle already mentioned that a little bit earlier is my hours that I can spend on this type of work are limited. And so I have been stretched pretty thin between all these conversations that are happening now in these meetings and I think it's really great. I think we've been making good progress in having these conversations, but it has impacted my time, for actually doing triage. So I think this is me raising the question of would it be okay for me for example to focus on doing this type of documentation and stuff like that for a little while and

00:35:00

Tim Krones: Kind of putting triage on the back burner or because I feel like I don't have enough time to do both Justice.

Michelle Philbrick: Tim I'm also happy to help maybe we can create a bones outline of something and…

Feanil Patel: But yeah.

Tim Krones: That makes sense.

Michelle Philbrick: then both work on filling it in.

Tim Krones: Sure.

Michelle Philbrick: Because I'm happy to do that as well.

Tim Krones: Yeah, and…

Adolfo Brandes: I raised my hands.

Feanil Patel: Adolfo

Tim Krones: I don't mean to exclude you from that. It's just I feel like I'm at a point now where I need to sort of prioritize and really be mindful of where I'm spending my time so that I can actually produce some. Useful work if that makes sense.

Feanil Patel: the Adolfo go for it.

Michelle Philbrick: Yep. Gotcha.

Adolfo Brandes: I'm volunteering to help with this very thing so Count Me In Michelle and

Feanil Patel: So I think the other thing I might suggest is even…

Tim Krones: Great.

Feanil Patel: though it's not a great long-term plan recruiting a third person for the contributions triage role. for the next six months pessimistically while we get the rest of this off the ground might be really useful Tim so that we can have more of your time spent on the Strategic stuff and less on the Tactical stuff so that you can spend more of that time in I think the place that makes the most sense so maybe

Feanil Patel: Maybe you and Michelle could think about it posting to discourse for looking for a third person. I know that there's a upfront cost of training that person on a process. We're trying to get rid of but I suspect it'll be around long enough that we will actually benefit from having that defraying of sort of capacity needs across one more person so that you guys could start work on some of the Strategic stuff for a little bit faster. So Even though it's not the direction. We want to go long-term. I think it's worth taking that step right now.

Feanil Patel: All And we're over my new half hour time slot. But does anybody have? Did you?

Robert Raposa: Really but every symptom and just a quick question on the process for that because if I had thoughts about SLA it's for maintainers and all kinds of things. does it make sense to make whatever that document is maybe being ADR that's underneath the maintainers. up or something so that everyone can it.

Feanil Patel: it'll probably be both I think there should be a document that's oriented towards getting people up to speed and whatever decisions are encoded in. It should be in an ADR underneath the maintainers and I can make sure that the ADR happens once the documentation that is sort of oriented towards training exists instead of the other way around but because I think we'll be able to change those sla's and some of those details in either direction. So it feels like getting something that's useful to people first feels like the right place to go. Kyle

Kyle McCormick: Are we missing any action items the only two I have nowhere. about python three points 12

Feanil Patel: Michelle can I ask you to try to recruit a third contributions triage person?

Michelle Philbrick: All Where's the best place to post that on the forums? Okay. Yeah.

Feanil Patel: Yeah on the discourse forums I think is the yeah.

Feanil Patel: And then Michelle and Tim it sounds a little bit an Adolfo. it sounds like starting I start writing up a maintainers. handbook almost

Adolfo Brandes: Yeah, That sounds good.

Jeremy Ristau: I thought I heard the other way as well.

Feanil Patel: and I'm sure there's

Jeremy Ristau: So maybe a PR template or something of content for submitters as well.

Feanil Patel: Yeah.

Adolfo Brandes: At the very least in a link to this document. Or something or the version of it or…

Feanil Patel: Yeah. Yeah.

Adolfo Brandes: meters, right?

Feanil Patel: You guys can reword this and split it up into whatever you want. But I think those are the things we're saying right now and we can check in on that in the future. Yeah, I think. If you guys come up with some words that it would be useful to tell every person. That makes a pull request. That's an open source contribution. It is pretty easy to update the bot to say those words to them.

00:40:00

Feanil Patel: So if that's the thing that you guys can do quickly we can make that change much faster and see how it's working.

Adolfo Brandes: so the difference is that the bot can identify whether that's sort of a drive-by contribution so we can have different texts than a PR template which would apply for everybody. Right? Is that the

Feanil Patel: The button identify a person that needs a CLA from a person that already has one. Okay, so I think that's…

Adolfo Brandes: Okay, Yeah, okay.

Feanil Patel: what I would use for now identifying a drive-by would require knowing how often that person is contributing and having a lot more sort of data back decision-making then it's capable of at the moment. But yeah.

Adolfo Brandes: No, just like the CLA is a good proxy.

Adolfo Brandes: I think. Yeah.

Feanil Patel: Yeah, I think it'll be good enough for…

Feanil Patel: what we need for Does anybody have anything else?

Michelle Philbrick: I just want to confirm it'll be documentation for both maintainers And contributors,…

Feanil Patel: and contributors Yeah.

Michelle Philbrick: I definitely work on stuff from the contributor side just…

Adolfo Brandes: already

Michelle Philbrick: and I'm sure that's a Tim can help on that as well because we see every day these come in.

Feanil Patel: Yeah, yeah.

Michelle Philbrick: The information that will be needed and everything.

Feanil Patel: and Kyle's just link to the existing how to maintain doc which is I think the thing to continue to improve adolfo's already started some of that, but I think you guys could great.

Adolfo Brandes: Yeah, exactly. That's what I was going to say. I figure they're a couple of more. Words we can add into that very PR regarding…

Feanil Patel: Yeah. Yeah,…

Adolfo Brandes: how to

Feanil Patel: yeah, I think flushing out that ongoing maintainers page and making it a lot more useful for maintainers and a lot more like clear to action will be helpful.

Feanil Patel: Big meeting. That's when I get for trying to shrink the meeting then. but anyway Yeah,…

Adolfo Brandes: useful one though wasn't waste and

Feanil Patel: yeah, definitely not wasted. I don't feel like I wasted the time I just feel bad for my next meeting which I told I'd be there. 15 minutes ago,…

Adolfo Brandes: oops Sorry.

Feanil Patel: but they understand they understand.

Feanil Patel: but Real quick. Yeah, I think we're good. I'm Gonna Keep it 30 minutes and let's see what happens next time. But if we hit we go over again, I'll just make it larger again and we'll adjust. Because it's easy to change things. I thank you everybody. Appreciate it.

Michelle Philbrick: Bye.

Adolfo Brandes: Bye.

Feanil Patel: Have a good day or night.

Kyle McCormick: 

Tim Krones: Thanks.

Meeting ended after 00:43:05 👋