2024-05-23 Meeting notes

 Date

May 23, 2024

 Participants

  • @Sarina Canelake

  • @Feanil Patel

  • @Jeremy Ristau

Previous TODOs

 Discussion topics

Time

Item

Presenter

Notes

Time

Item

Presenter

Notes

 

Updates

@Feanil Patel

  • Switching edx-platform unit tests to standard ubuntu runners

  • CI will fail if a Python package is added that lives in the edx org

    • Encourage EXTRA_REQUIREMENTS instead of direct deps

    • Not just edx. It’ll be a denylist

      • edx

      • mit-odl

      • open-craft

      • et al

    • Third-parties (Django) are fine

    • Temp allowlist for current ones



High-level arch brainstorming/tracking - what’s a good space?

@Kyle McCormick

https://docs.google.com/document/d/1s-Aqbq8ohA-YG9jhrV3VCQyAUXDQgE8nq4DH7Ntr79E/edit

Where should stuff like this ^ live?

edx-platform GitHub issues

Board?

  • Feanil: maybe not

  • Robert: how do we prioritize?

Docs for things we’ve already decided - e.g. no new legacy frontends

Put it on the platform-roadmap? when it’s ready

Maintenance board?

Apply labels:

  • code-health + epic

  • architecture + epic ?

Kyle will turn the google doc into issues and figure out how to organize them

 

edx-platform app-level docs

@Robert Raposa

Ideas:

  • Add app-level READMEs

  • Add top-level index page to make discovery easier.

  • Eventually update OEP-19 with best practices for exposing these for a repo.

 

Redwood bug triage for edx-platform

@Kyle McCormick

We expect that

  • bugs will be found in release testing

  • BTR will triage them

    • if they belong to an active dev team (eg tagging), they work on the fix

    • but if they are unexpected regressions, then BTR will triage the severity..

      • if it’s a release blocker, then BTR will find the maintiner

      • if it’s edx-platform, then we are the maintainer, so we’ll need to figure out how to find the right person/team to do the fix

      • Let’s cross that bridge when it happens. No action needed now.

[switch to full maintenance WG mode]

3m

 

Sarina

Any appetite to update guidelines around pinning constraints? See idea: https://github.com/openedx/open-edx-proposals/issues/332

  • Please provide your input on this issue.

 

Dependabot PRs

@Robert Raposa (unless @Jillian Vogel joins)

From @Jillian Vogel:

I've started maintaining a couple of openedx repos, and am working through the dependabot PRs. 

Does anyone have any advice for me?  I'm curious how people manage the continual stream.

  • For minor version updates, is it OK to trust the CI, or are people doing manual testing too?

    • No, trust your CI but if you don’t trust your CI, invest in it.

  • When to bump the package version?

    • Updates should run at-least on a weekly basis. One recommendation is that you make all the schedules for the repos you maintain, the same.

    • Dropping support for major upstream deps(django, drf, react) should be done in coordination with the rest of the edx-platform project.

    • Updates to test requirements don’t warrant a version bump or changelog entry, since they are not exposed by setup.py

  • Docs that could be updated :

 

Un maintained repos

@Chintan Joshi

- Chintan is gonna try to do some of the initial maintenance on a bunch of these.

 Action items

@Kyle McCormick turn the Big Ideas google doc into issues and figure out how to organize them
@Robert Raposa Add edx-platform issue for improving the docs (see notes above).
@Chintan Joshi Update maintenance docs with guidance for when to merge regular dependency update PRs

Recording and Transcript

Recording:

edx-platform Maintenance  Sub-Group (2024-05-23 09:03 GMT-4) - Transcript

Attendees

Adolfo Brandes, Awais Qureshi, Chintan Joshi, Feanil Patel, Feanil Patel's Presentation, Jeremy Ristau, Kyle McCormick, Maksim Sokolskiy, Maria Grimaldi, Michelle Philbrick, Piotr Surowiec, Robert Raposa, Sarina Canelake, Tim Krones

Transcript

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

Feanil Patel: and communicating

Robert Raposa: Yes, and with that. Are we thinking also maybe in edx platform project or something that can be used to? Manage the issues no?

00:05:00

Feanil Patel: I think that's less valuable unless there was an edx platform team that was working off of that project and

Robert Raposa: So how does one see prioritization of these things or it's just individual teams or organizations that choose their own priority and pick the issues they want to Pick us.

Feanil Patel: right Yeah,…

Feanil Patel: yeah exactly. There's don't the things that are important to different teams and groups might Be different. I think we want to be clear about sort of directional if you start trying to add a third. High Level app to edx platform we will probably stop you. If you try adding lots of Legacy front ends. We will probably stop Here's how far the Legacy front end removal is so far and if it's important to you just have only mfes. here's the work left to do. Does that sound right Kyle?

Robert Raposa: I'm gonna

Kyle McCormick: That's what you just said. Sounds more like contributing guidelines. Which seems like something we should pen into rst.

Feanil Patel: Mm-hmm

Feanil Patel: Okay.

Kyle McCormick: I like it as developer facing documentation like things we've kind of already decided I don't think it's controversial that we shouldn't be adding more Legacy front ends. we don't need to track them and take it we just Put it in at next platform.

Kyle McCormick: If we're talking what's an example from this talk? Whittling down the base requirements list. That's when we actively have to do or it's never gonna happen. And it's also a long-term thing with multiple approaches we could take So you think like a GitHub issue?

Feanil Patel: right

Kyle McCormick: that And maybe put a label on it. So this group could keep track of those sort of ideas.

Feanil Patel: Yeah, because I'm thinking, in terms of visibility and I'm totally okay with a project board. If we think that that would be a valuable way of organizing I just like

Feanil Patel: My current reaction is I already have enough project boards, and I don't need another one unless there's a really good reason for it is my default. So I'm happy to be wrong about that. But I want people to make the case for that project board and why it would be different from other ones my one thought when you said that is that these probably will need to be on the edx platform roadmap. Because they will be significant enough architectural changes that even if there's no user facing change. We want that communicated to the community.

Kyle McCormick: Sure, does the platform Road app have space for things that are really Half Baked ideas?

Feanil Patel: In the same sort of vein as the major upgrades and any new feature additions at named release time.

Feanil Patel: which it's fine for it to be an edx platform issue and then we connect it with the platformer roadmap project that already exists.

Kyle McCormick: I agree that a board might be too heavy to start. Could I propose just a label that way we can keep? track of these somehow Maybe

Feanil Patel: so this is the nice thing about it being an issue is the issues can be half baked and when they're about to be implemented we add them to the platform roadmap for when we think we are going to be working on them.

Robert Raposa: And sorry when you're proposing label can you propose an example label name? Just so I understand the scope…

Kyle McCormick: yeah.

Feanil Patel: Okay. yeah,…

Robert Raposa: what things you're talking about.

Kyle McCormick: I mean,…

Feanil Patel: we can also add them to the maintenance board and…

Kyle McCormick: there's the code Health label already.

Feanil Patel: just filter them out and…

Kyle McCormick: It's meant for refactorings and such.

Feanil Patel: have a

Feanil Patel: Separate view of the maintenance board.

Kyle McCormick: maybe that plus ideation

Feanil Patel: That's Alex platform specific if that would be helpful.

Kyle McCormick: discovery

Robert Raposa: You got Just something of that nature. Okay, I just wasn't clear what you meant and now I understand.

Robert Raposa: We could add that to the notes Here.

00:10:00

Kyle McCormick: maybe just code health and epic and I think those together indicate this sort of things that this document.

Feanil Patel: Yeah, let's platform. I was like a ton of labels right now.

Kyle McCormick: is looking for

Feanil Patel: including a bunch that Not sure what the value is…

Kyle McCormick: just

Feanil Patel: but aren't necessarily an important thing to fix today.

Kyle McCormick: Alright, there's also the maintenance label.

Feanil Patel: architecture might be another label that It'd be good to add to some of these…

Kyle McCormick: all What I can do is start turning this doc into issues.

Feanil Patel: but is not indicative of the things that you. because in terms of reorganizing sort of The top level holders or…

Kyle McCormick: And then I think it'll become. apparent but we should label them and I'll share this with you guys.

Feanil Patel: removing mongodb.

Kyle McCormick: Yeah, I just want to have some way to filter for them so that I can point instead of Link I can link people to a set of issues that is communicating the same thing.

Feanil Patel: That sounds good.

Robert Raposa: Yeah, just a quick note and I'm …

Feanil Patel: All right.

Robert Raposa: again super glad that you made progress and got the Ducks available. So, I don't know if I can add a link and help. People see what I'm about to talk about. But let me

Feanil Patel: cool Robert,…

Robert Raposa: yeah, but basically there's lots of dark directories within apps that have their own decisions and…

Feanil Patel: chat about the app level Docs.

Robert Raposa: they're now in the dark or in read the docs output, but they are. difficult to find so

Feanil Patel: Yeah, I can find it if you want to keep talking.

Robert Raposa: just a possibility of maybe improving that with a different index page or something that helps

Robert Raposa: help bring them to let

Feanil Patel: right Yeah. Yeah, I mean this is sort of generated from the current. Yeah, I think that's a great idea. So, let me put a link to this. top level

Robert Raposa: It sorry when you say they didn't put them in here. Do you mean in the docs directories or…

Feanil Patel: Yeah, and one of the decisions that team made initially that I'm sort of.

Robert Raposa: just in the output of read the docs? Yeah, yeah agreed.

Feanil Patel: Curious about other people's thoughts on is that they currently did not put all the readme's in there.

Robert Raposa: I think it's a definite duck.

Feanil Patel: And I almost wonder if it would be useful to have the read maze that are at the app level. In here just to sort of better Orient people.

Robert Raposa: Yeah.

Feanil Patel: in the output. Yeah.

Robert Raposa: Okay.

Feanil Patel: Yeah, yeah. I think they were thinking those are sort of more relevant to people browsing GitHub Than People browsing the docs,…

Robert Raposa: Are really maintenance.

Feanil Patel: but I think that You kind of want that orientation either way.

Robert Raposa: I guess this could be a GitHub issue.

Feanil Patel: Yeah.

Robert Raposa: I guess just collects this information.

Feanil Patel: Don't know when they'll get back to work on it. So I think the next iteration will probably be a little while unless somebody else has capacity to work on it sooner.

Kyle McCormick: To have an idea of how fsible even desirable would be to merge the doc's tree and the python doc strings tree together.

Feanil Patel: Yeah, I think this is a great GitHub issue for an enhancement to the top generation.

00:15:00

Robert Raposa: Thank you.

Feanil Patel: Yeah, they tried to do that dock strings tree is generated by a different plugin right now. And so it was difficult to sort of overlay the two. On top of each other in a same way. I think that they tried it and it didn't make super much sense to…

Kyle McCormick: cool

Feanil Patel: I mean, there's the plug-in for generating. These other docs is now in edx platform.

Kyle McCormick: another question,…

Feanil Patel: So there's definitely a bunch of enhancements we could make to make Sort of discoverability and…

Kyle McCormick: how do you see this fitting into the Navigation experience of docs that open at .

Feanil Patel: browsability better.

Robert Raposa: Could be nice.

Feanil Patel: But at least we've now proven out that we can get at those docs which previously we were not able to access by using the default the Sphinx tooling. So yeah.

Feanil Patel: So right now it's part of the edx platform reference documentation.

Kyle McCormick: So it does have Concepts and…

Kyle McCormick: how to use.

Feanil Patel: so the next platform documentation is currently getting generated whenever we merge to master and…

Kyle McCormick: and references itself being underneath references in edx platform

Feanil Patel: that's linked from the developer reference documentation on Dockside or So you'll see it's all searchable under that one domain. and so Should hopefully be fairly discoverable.

Kyle McCormick: right

Feanil Patel: Yeah. Yeah.

Feanil Patel: Each this and this is an organization complexity with the fact that we have so many repos. Is that one of the news early decisions was like each repo should have its own sort of quad.

Kyle McCormick: Yeah. Okay, cool. I feel like we could talk for a long time about.

Feanil Patel: Even though it the list of all of the repos might live on the references in the top level.

Kyle McCormick: This is pretty interesting.

Feanil Patel: But if they all have their own quad, we could potentially build out New table of contents that are like here all of the concepts from across the system or…

Kyle McCormick: Okay.

Feanil Patel: here's all the reference stocks from across the system.

Kyle McCormick: Yes.

Feanil Patel: yeah, there's like yeah,…

Kyle McCormick: yeah, we all start with search anyway,…

Feanil Patel: there's a bunch of other ways to organize this kind of With the amount of time I have for it that sort of primary goal has been to get as many things under this one domain as possible so that we can now do Google Search.

Kyle McCormick: right

Kyle McCormick: over generation I know but Cool last question.

Robert Raposa: Yeah.

Kyle McCormick: Are you thinking that the specific stocks like this should be and operator Focus rather

Feanil Patel: For one domain and find all the docs.

Feanil Patel: And that sort of enables a lot more Discovery than any amount of tree structuring that I can do.

Kyle McCormick: Okay.

Feanil Patel: Yeah, exactly.

Feanil Patel: but

Feanil Patel: Yeah, no, Isis. Yeah, I suspect that some especially the things that are user facing bits will use your facing documentation as well.

Feanil Patel: Right. at least the educator side of the world wants to build that all in one place rather than across all of the repos. So that is sort of being lived like Loops under the educator persona. And that kind of makes sense because they don't care that this is coming out of a plug-in or not. They just want to know by default. This thing is enabled. and if it is not You should hear operator instructions for…

Kyle McCormick: Yeah.

Feanil Patel: how to enable it which might involve installing the Plugin or…

Kyle McCormick: That sounds good to me. especially since features cut across repos usually

Feanil Patel: not. So I think

Feanil Patel: They could have other persona docs under there.

Robert Raposa: so I added an action item for myself to create the edx platform doc ticket

Feanil Patel: I think they won't in the short term just because I think that we're currently building that high level dock of what's available by default and…

Kyle McCormick: Thanks.

Feanil Patel: what's not.

Kyle McCormick: No. Probably is worth this Cafe.

Feanil Patel: yeah, yeah,

Kyle McCormick: Thank you for Being that up. I feel like have you heard of any next platform bugs coming out of release testing yet?

Feanil Patel: Okay, awesome.

Kyle McCormick: Because I haven't

Feanil Patel: they're the one around reruns not being available, which was unclear if it was a front-end bug or it was a studio front end bug or if it was a platform backend issue

00:20:00

Jeremy Ristau: And when we say Redwood bugs,…

Feanil Patel: I believe. It's an accent improvements is looking at that one right now.

Jeremy Ristau: we're talking about bugs that would block the release right? We're not talking about all of the Tagging bugs that Jenna has noticed and…

Feanil Patel: But they've been having some trouble trying to reproduce it.

Jeremy Ristau: her testing because we're trying to get that UI out.

Feanil Patel: I haven't seen any sort of like.

Jeremy Ristau: We're talking about things that are blockers, right?

Feanil Patel: Official bug tickets from BTR yet. This was one that filed. so we're talking. Yeah, go ahead.

Feanil Patel: 

Kyle McCormick: It's I think things that come up.

Feanil Patel: As he was doing his testing.

Kyle McCormick: From release testing that aren't expressly being handled by a Dev team. So there's a tagging team and their job is to get tagging working in Redwood. But if we find some completely unrelated bug in the next platform. then

Kyle McCormick: Sorry that emoji distracting me so much if there's a good.

Robert Raposa: That's right.

Kyle McCormick: If we have some unrelated bug another platform. Someone will need to decide if it's released blocking. I think BTR does that and if it is then I think they would. Say we need someone to fix this and I think that's where. We would come in.

Jeremy Ristau: So it's less the triage and more the picking up of the eyes work.

Robert Raposa: perfect back

Kyle McCormick: I think so.

Jeremy Ristau: And doing the work.

Kyle McCormick: Yeah, right BTR history edgers. So I think we don't need to be doing that.

Feanil Patel: Yeah. Yeah, they'll determine what is a non-blocking bug along with product working group and at some point it may just be that they need. Maintenance for specific repo to go in and…

Robert Raposa: quality

Feanil Patel: actually try and fix it.

Kyle McCormick: Maybe this would be easy to talk about when one happens when we get a bug from BTR and then we have to figure out what to do. I guess I'm imagining that there will be some conversation along the lines of Here's this bug. What team is supposed to fix it? Can they fix it? Do we need some backstop to fix it? And Yeah, the sort of things will be talking about.

Jeremy Ristau: Yeah.

Jeremy Ristau: And BTR would still be driving that. conversation

Kyle McCormick: I think good question. I think that BTR would be

Feanil Patel: BTR I think is gonna get to the point of here is the issue that is a blocker for the Redwood release inform maintainers at that point maintenance. I think would drive we need somebody to fix it. What is available in our sort of resource pool who can take this on who's the right person to take this on and…

Jeremy Ristau: Yeah, okay.

Feanil Patel: like getting it assigned out to the right team and making sure it actually gets done.

Jeremy Ristau: So it's really? Very conversation that doesn't have anything to do with the maintenance working group. Unless a specific maintenance says, I can't get to this. Let me turn to my resources. One of my courses is other maintenance.

Feanil Patel: yeah, yeah, but in particular I think Kyle is bringing this this is not a maintenance working group problem, but a edx platform maintenance problem as Alex platform maintainers might run into this issue for edx platform and…

Kyle McCormick: right

Feanil Patel: we should be prepared to Have time for it. And be able to find the right resources if there's expertise either add to you or an open craft or wherever that we need to track down.

Jeremy Ristau: Gotcha.

Feanil Patel: Yeah from the maintenance working group perspective. I think. We hopefully will not touch the release too much. I think our goal is to get things into the release and now that they're in there. They've got it from here.

Jeremy Ristau: Right, that makes sense.

Robert Raposa: by the way, that reminded me I don't think we have time for this now, but at some point we should probably come back to the topic of we as edx platform maintainers and through that we is and what it means and we're where that issue is going so Okay.

Feanil Patel: yeah, yeah. Forza I think my based on our conversations last time, I think I'm leaning towards the four of us are the platform maintainers and we will sort of operate from there and just do the work and see how it goes.

Robert Raposa: Good, And that mean you'll be making updates that yeah.

Feanil Patel: Yeah, I'll update that PR. Yeah, yeah.

Robert Raposa: Okay. cool Yeah,…

Feanil Patel: I just have not had time.

Robert Raposa: no worries. Just didn't know where you were…

Feanil Patel: Yeah, yeah,…

Robert Raposa: where you're thinking was.

Feanil Patel: that's where my thinking is right now is I based on the conversations we had last time. I think it's like let's do that and then We'll sort of move from there.

00:25:00

Robert Raposa: Okay. Thanks.

Jeremy Ristau: Person personally May appreciate you putting the 10K a month. Runners problem above right I personally appreciate that.

Feanil Patel: right Right.

Feanil Patel: I like this is a thing, but it's not the thing right now.

Jeremy Ristau: it's one of

Jeremy Ristau: Yeah.

Feanil Patel: Plus it's the sort of thing where it's Give background time is really what it needs. Not a lot of foreground time. And occasional conversations are useful.

Feanil Patel: All right.

Robert Raposa: Okay, Kyle or female? I don't know. If one of you want to update the notes part of the robot bug triage section. Or whatever.

Kyle McCormick: Yeah, I can add a little summary.

Feanil Patel: yeah. Yeah. Thanks Kyle.

Robert Raposa: Thanks. Thanks.

Feanil Patel: and then we can talk about. the dependabot jail question Jill probably won't be doing since it's like

Feanil Patel: one in the morning or something in Australia.

Robert Raposa: Yeah. Yeah, I figured now but

Feanil Patel: 11:30 stop.

Robert Raposa: I don't actually need to resent this. I just was adding a name.

Feanil Patel: Yeah.

Robert Raposa: So feel free to switch that to finale.

Feanil Patel: Yeah.

Feanil Patel: I think that these are good questions and I have an They raised the question of where could that answer live such that I would be able to not have to say it as often,

Feanil Patel: But I can reply to Jill's original slack as well with that and maybe she has some ideas.

Jeremy Ristau: anything you Mind zooming in just a click or…

Robert Raposa: underwear

Feanil Patel: Yeah. …

Jeremy Ristau: two. Thanks.

Feanil Patel: yeah, absolutely.

Feanil Patel: Yeah, I've got the whole screen today. So I can also make it wider because

Feanil Patel: it doesn't believe in you. And at the table is not wider. It doesn't make sense.

Feanil Patel: this better

Feanil Patel: this whole internet thing. I'm still not too sure about.

Feanil Patel: It was fine when it was all text. But all of this styling is really

Feanil Patel: I don't know if it's made it better,

Kyle McCormick: Are we gonna be working on HTML only mode for edx platform? Is that what you're saying? Anything else?

Feanil Patel: edx platform telnet Edition just like Tell that X platform.

Kyle McCormick: awesome

Feanil Patel: it's got an AI chat interface and then you just Talk to it. So you learn stuff?

Kyle McCormick: big ideas for the next hackathon

Feanil Patel: Yeah. there's images, but they're all ASCII art.

Feanil Patel: Pickup how much less distraction there would be. in their education Yeah,…

Kyle McCormick: the front end build time would be No, great.

Feanil Patel: exactly. Yeah.

Feanil Patel: These are the ideas. They don't want you to hear.

Feanil Patel: You can tell it in my sleep an extra punchy.

Feanil Patel: Okay, Sarina, you have the first topic you want to kick us off.

Sarina Canelake: I think this is short. I've been just going through All the issues and open pull requests on the oep repo as part of my own maintainership. this is an idea that was floated that I don't have an opinion on but I would like the people in this group to give their opinion on around basically dating in files when python dependencies were last upgraded.

00:30:00

Sarina Canelake: I don't know if I like this might be helpful, but it might not be because we can always just blame the file right so Yeah, I think…

Feanil Patel: Yeah.

Sarina Canelake: if you have an opinion. Maybe just say it on this I'd rather capture this conversation on this issue. I'm happy to close it if it's not a good idea.

Robert Raposa: And just to clarify work over speaking specifically about the constraints file not all files just constraints.

Feanil Patel: right

Sarina Canelake: Yeah, thank you Robert for that clarification.

Robert Raposa: nurse

Feanil Patel: Yeah, that sounds Cool. Yeah, so people should take a look at that after the meeting and provide some feedback. Think especially of if it's in your repo. How would you ensure that? It How would you make sure what would you do? Are you okay with having this tickets existence Etc as maintenance?

Feanil Patel: cool and then Jill asked a very good question about in slack which was around sort of like How do you keep up with dependabot PRS and if you have front end repos also renovate PRS?

Feanil Patel: I think we generally have those automations running on a weekly basis.

Feanil Patel: I think one of the big recommendations that I would have is to make that Weekly Cadence for all of the repos you maintain be the same. It is a lot easier to review all of the update PRS on a Thursday morning when they have quiet time, then one on Monday and another one on Tuesday and then two on Wednesday and then three on Thursday.

Feanil Patel: we've done this for a lot of the ones that accent maintains all of those PRS get generated on Sunday evening so that Monday morning we can just review all the dependency updates and then do them. We believe a lot of this is captured in the original maintenance. Docs I'll see if I can dig those up after the meeting and post them but the motivators behind sort of doing this on a weekly basis rather than sort of on a wider Cadence was that this is also how we get security updates on our requirements and we want to capture those as quickly as possible in master.

Feanil Patel: but in terms of sort of whether or not to sort of go to this first question of do you need to be doing manual testing? The answers know if you think that something requires an annual testing because it especially for my minor version. You should just be able to use UCI. and this part is personal opinion if you can't get your CI to a point where you can trust these things we should be talking about making your CI better not about adding more manual testing to sort of libraries and small bits of our code. So let's

Feanil Patel: the goal should be that these are really low hanging fruit where either they are green and you can just press the merge button or they are red and you need to look into why they are red and either fix your test fix the parts of that system that you rely on that library that you rely on.

Robert Raposa: so I guess the note there's Is being updated?

Feanil Patel: What's that Robert?

Robert Raposa: I was going to say exactly what someone just typed.

Feanil Patel: cool

Kyle McCormick: Is anyone willing to add this information to the maintenance?

00:35:00

Robert Raposa: By this second question in terms of Kyle but we'll see if anyone actually speaks to say that they will do that. But I'm just wondering about the second question where I'm seeing stuff type do we think that the meaning my interpretation of this is if I have a repo that is a library. And I'm updating dependencies. Yep. Or what types of updates should I also bump the version of my library?

Robert Raposa: for example if we were Dropping support for a major version or something like that. So

Robert Raposa: and I still hope someone volunteers for what Kyle was asking.

Feanil Patel: right because most of the libraries are pulling in the dot in files as their Package requires which don't usually have version pinning in them.

Feanil Patel: this is probably a space where actually something enhancements could be made because It's one of these things that could lead us into trouble but I think because most things are updating reasonably regularly it is fine, which is that because we're only pinning the package name and not specific versions of it. it is theoretically possible that major version bumps happen where you make changes to a library to only be compatible with the latest major version. But the Upstream does not capture that doesn't have that major version yet. And so it breaks things Upstream when they try to run it.

Feanil Patel: I don't know if it's the problem today, but if people have sort of thoughts and feelings on that or potential enhancements that aren't significantly more effort. I would love to hear about that.

Feanil Patel: I think we sort of manage this for the big dependencies like Django and drf. Via constraints right now where we don't bump major versions of those without sort of coordinating across the entire ecosystem. but you could imagine this happening with something minor like

Feanil Patel: I don't know some third party library that everybody uses requests has a major version and they change their API fully or something.

Feanil Patel: unlikely, but

Kyle McCormick: This happens to her we do this to ourselves too. We're like for example, if a breaking change is made to the Xbox repo and then we run make upgrading our next platform.

Feanil Patel: right

Kyle McCormick: then make upgraded broken until edx platform is adapted to deal with that breaking change

Feanil Patel: right

Kyle McCormick: right

00:40:00

Feanil Patel: I think that one of the things about things like Paragon, is that on the Jazz side? It's a little bit nicer because the json file tends to have more pins in it. historically we haven't been doing very many pins in the infiles, but we could be pinning to Major versions in the infiles or something like that. I'm not sure that our upgrade automation handles that in a way where it doesn't increase our workload because I think there's no automatic way to bump the pins in file. Whereas I think with renovate it will handle bumping the pins in the package.json file. always

Feanil Patel: right

Feanil Patel: right But yeah, so we could be Auto emerging more of these PRS. I don't think that helps us with the libraries still in file when they declare their dependencies instead of the text file, so The libraries request extra requires don't look like Is this python innards right? It's like the way you declare dependencies in Python doesn't the pins currently in edx platform. So I think we'd have to do that something like that to make better use of this.

Feanil Patel: Robert

Feanil Patel: yeah.

Feanil Patel: I would personally love a volunteer from the maintenance working group. If we can muster one.

Feanil Patel: how

Kyle McCormick: Going depends. I do want to make sure we're not over engineering. I think when we have the makeup grade process happening where there's a make upgrade PRS on every repo and maintainers are regularly emerging My experiences that it works great. And that people look at the diff and they see a major version bump and that's a signal for them to be like maybe I need to look a little closer at what packages change what changes here and if it's all batch version, but instead it's like just merge don't even think about it. if we are finding that that's not working then I think maybe looking into Stuff with semantic versioning could be good. But I think we just need to get back on the upgrade treadmill first.

Feanil Patel: Chintan, did you have your hand up?

Chintan Joshi: Of yeah, I was seeing for volunteering with postings. Anyways learning so with learning I can help a little.

Feanil Patel: But yeah awesome.

Feanil Patel: Yeah, that sounds great. If you run into any issues, feel free to hang in the maintenance channel on I'm sure we can. support you

Feanil Patel: Add it on the list. There's two of you.

Feanil Patel: Pick a lot of the picture.

Chintan Joshi: I'm the one without the picture.

Feanil Patel: 

Feanil Patel: Is that just a different person entirely?

Chintan Joshi: I think it.

00:45:00

Feanil Patel: Cool. anybody have anything else to cover today?

Chintan Joshi: I had one more thing you asked me to look for repose which doesn't have any maintenance and isn't updated a lot. So From that I've created a sheet. I'm still updating shred it in the job.

Feanil Patel: Okay, that's good. Yeah, I've been looking through this and sort of following back up with people who have sort of received access but have not yet sort of updated the config or started doing their work. I think a question for this group sort of longer term is going to be like once we have maintenance understanding how well things are being maintained. It's sort of the next question down the line. We're definitely nowhere near having maintenance for everything yet, but I think we're getting closer to Asking that question of and this is going to be impossible to read. So let me just drop it in chat. You guys can open it up.

Feanil Patel: Because if I make it bigger, I can't see all the columns to discuss anything.

Feanil Patel: But yeah, once we have maintenance I think sort of seeing how often requirement update PRS are happening and if

Feanil Patel: there are any Other issues that we need to create across the board or things like that.

Feanil Patel: A bit Yeah, if there's a bunch of repos definitely that's still need to maintainers. So those are ones that if you have a list that you want to sort of start working on, that'd be great.

Feanil Patel: So you've got this?

Feanil Patel: nice, okay.

Feanil Patel: So you're gonna work on these ones here?

Chintan Joshi: Yeah, I'm still preparing the list which are Repose you mentioned that I can take up those proposal. I'll take them.

Feanil Patel: Okay for adding in the JS dependencies tooling you should be able to do that from a fork even so you don't need right access to make those pull requests. So you could even start by doing that across a bunch of these.

Chintan Joshi: Okay.

Feanil Patel: All right. other questions thoughts feedback

Kyle McCormick: Chintan, I noted that a couple of those are differed. just so you don't put a bunch of effort into them.

Feanil Patel: Okay. I think that's everything folks.

Feanil Patel: Alright, see you next time.

Kyle McCormick: Thanks, everyone.

Meeting ended after 00:48:56 👋