2024-08-22 Meeting notes
- Feanil Patel
- Robert Raposa
All public Working Group meetings follow the Recording Policy for Open edX Meetings
Date
Aug 22, 2024
Participants
@Feanil Patel
@Michelle Philbrick
@Adolfo Brandes
@Kyle McCormick
@Sarina Canelake
@Awais Qureshi
@Maksim Sokolskiy
Previous TODOs
Discussion topicsT
Item | Presenter | Notes |
---|
Item | Presenter | Notes |
---|---|---|
Ubuntu 20.04 → | @Feanil Patel |
|
Nascent proposal for getting maintainer involvement during the release testing/fixing window | Adolfo | Roughly:
Feedback:
|
T&L Frontend-content-components repo | Michelle |
|
ecommerce Deprecation | Feanil | |
--- edx-platform topics below here --- | ||
xblock work |
|
|
Resiliency |
|
Action items
ubuntu-latest
for CIRecording and Transcript
Recording:
Maintenance Working Group Meeting – 2024/08/22 08:59 EDT – Transcript
Attendees
Adolfo Brandes, Awais Qureshi, Feanil Patel, Feanil Patel's Presentation, Kyle McCormick, Kyle McCormick's Presentation, Maksim Sokolskiy, Michelle Philbrick, Piotr Surowiec, Robert Raposa, Sarina Canelake
Transcript
Robert Raposa: simulator tab
Kyle McCormick: Hey Robert.
Robert Raposa: I'm going to step away for about 30 seconds or so. I'll be right back.
Kyle McCormick: That's good.
Adolfo Brandes: Hello.
Kyle McCormick: Morning, Adolfo.
Adolfo Brandes: Kyle when you rubber
Kyle McCormick: Forever says he'll be back in just a minute.
Feanil Patel: If they have pretty fast meeting today, I think.
Adolfo Brandes: that's cool.
Michelle Philbrick: Lola
Adolfo Brandes: She here.
Kyle McCormick: Nice catch Michelle just the ear in the corner. good cat
Michelle Philbrick: I'm always on the lookout.
Feanil Patel: events
Robert Raposa: practices
Feanil Patel: But it share mice really.
Robert Raposa: it's
Feanil Patel: We're gonna have a lot more banter on these recordings now that it starts auto recording.
Kyle McCormick: Yeah.
Adolfo Brandes: There's always a drawback.
Michelle Philbrick: mostly
Kyle McCormick: sure, there's
Feanil Patel: I think it's okay.
Kyle McCormick: I'm sure there's an AI tool that can identify the actual start of the meeting from for us.
Feanil Patel: The meeting is now starting.
Adolfo Brandes: Get a clacker.
Feanil Patel: All right, good morning afternoon and evening everybody. Welcome to the maintenance working group. Let's get started with the usual checking back in on previous to news. We've got a topic for next time for edx platform. I believe that was
Feanil Patel: so I'll just add that down here.
Feanil Patel: and then
Adolfo Brandes: I have a BTR thing.
Feanil Patel: okay.
Feanil Patel: that means yes looking at the and then I have
Feanil Patel: maybe it won't be it with meeting. Yeah, maybe I should say that every time.
Robert Raposa: Mine might be at X platform related. I'm not sure but I'm the ex-block work that's going on.
Feanil Patel: Yeah.
Robert Raposa: I just wanted to check in with that and that's some questions and stuff.
Feanil Patel: All right.
Feanil Patel: All right, I'll start off first. I've been looking at the Ubuntu upgrade across our system and in particularly, sort of after this discussions last week as I was looking through that I asked myself seriously whether it made sense to Try to jump to 2204 or 20 jump to a targeted named release or use github's provided Ubuntu latest, which is a tag that I assume they will move. in the future to newer releases I think the only place I could come up where it made sense not to do that was in code jail where we want to test explicitly with specific Versions of Ubuntu and in particular app armor, it's the one that depends most on the underlying system.
00:05:00
Feanil Patel: For the rest, I believe any Ubuntu that can install the relevant system packages which are only relevant for edx platform and our documented in that repo. Should work with any Ubuntu so that was just thinking we would update our testing to test with latest and therefore reduce our future test burden by not having to do this again in the future. So
Feanil Patel: I am mentioning it here in case you guys have feedback. I'm gonna update the ticket to that effect and then I do need help. I think making that change across all of the repos. We find the maintenance ticket. And I'll link that here as well. I did make the underlying. issues or all of the repos
Feanil Patel: so that we can track what needs to be done. So that's all in this ticket that I've just linked.
Feanil Patel: And that's everywhere. We're 2404 is mentioned or 22 204 is mentioned.
Feanil Patel: so Fort Robert
Robert Raposa: and I think the two questions I have one Should there be exceptions and should Alex platform with code jail being exception or is it not really affect anything to I guess It depends, where we're using it within that X platform and…
Feanil Patel: Yeah.
Robert Raposa: maybe for the tests It's fine.
Robert Raposa: so it's a question.
Robert Raposa: and the other this is A related question. I was wondering about this ticket Is there a Decker and should there be a Deborah mostly even if there's not I mean just mostly around dates. I know you had. an interest having changes made at one point time, obviously if we move to Ubuntu latest, you'll have that concern once and…
Feanil Patel: Yeah. Yeah, That's kind of my thought.
Feanil Patel: Yeah, My current thinking was that we would Dipper Ubuntu 2004 once and…
Robert Raposa: Maybe never again.
Robert Raposa: I don't know so.
Robert Raposa: performing
Feanil Patel: moved to latest and that deprecation it would have essentially the things I've just said in it.
Feanil Patel: as rationale and then
Feanil Patel: for any sea for any CI that uses 2004 today we would have 2004 and…
Robert Raposa: okay, and go ahead.
Feanil Patel: Ubuntu latest testing for six months
Feanil Patel: and then after that it would just drop support for 2004
Robert Raposa: All right, because you had mentioned something about the dates and how you didn't you were hoping that it would be less than six months.
Feanil Patel: So yeah. Yeah.
Robert Raposa: And so it would just be great to have your dates and…
Feanil Patel: We can comment there. But yeah,…
Robert Raposa: wishes some place so that I could attach it to
Feanil Patel: I can do that. That I think I always wanted to be shorter,…
Robert Raposa: okay.
Feanil Patel: but I'm sure I think we'll just say six months and if there's an issue the reason that I remember the detail is that there are lots of workflows where it doesn't make sense to run them twice. And for those we'll just move to the new. Tube into this part publishing the i Pi linting and a couple of other things that aren't dependent on the underlying python 0.2 version. I think we should just move. And I'll mention that in the deprecation ticket so that the discussion can be had up there. But yeah,…
Robert Raposa: yeah.
Feanil Patel: thank you for that around here.
Robert Raposa: But I thought you also mentioned that you wanted it in on a named release. It was gonna hit before six months or something of that nature because of when it Actually,…
Feanil Patel: That's more a question for tutor.
00:10:00
Robert Raposa: but I don't know if that's true.
Feanil Patel: I think tutor should be running on. 22 or 2404 before Sumac because of when 2004 goes out of support. but That shouldn't affect any Master operators like yourself.
Robert Raposa: Okay, but is it?
Robert Raposa: You got it, okay. Sounds good. Thank you.
Feanil Patel: Yeah, absolutely. we will basically have September to get all of this landed.
Feanil Patel: Yes.
Feanil Patel: right
Feanil Patel: Yeah.
Feanil Patel: Yeah.
Feanil Patel: It is.
Feanil Patel: And to be clear that's not work that needs to happen by operators. Great. Yeah, yeah.
Feanil Patel: Okay.
Feanil Patel: Yeah, it's definitely a short timeline for maintenance and in the future. Hopefully if we just do it the latest we don't have to do this again, and that just reduces overall maintenance Burton for everybody.
Feanil Patel: It's kind of I thinking. But I'll capture all that in that part ticket. this week
Feanil Patel: Okay.
Feanil Patel: Adolfo you want to go on the next topic?
Adolfo Brandes:
Adolfo Brandes: Sure also I just wanted the tutor situation remember they were considering moving to the python Upstream image, but apparently that's longer a target for sumac.
Feanil Patel: Yeah.
Feanil Patel: Yeah, Regis came by last maintenance working group.
Adolfo Brandes: Okay. great.
Feanil Patel: We had a chat about it. Yeah.
Adolfo Brandes: Yeah, I invited them but I couldn't make it last time so.
Feanil Patel: No, he was here and we talked through it and pointed out that code jail is a critical requirement.
Adolfo Brandes: Yeah.
Feanil Patel: So it was happy it made a lot of sense to him and he was happy that it's a sort of revisit that later.
Adolfo Brandes: Okay, great. All right.
Feanil Patel: Yeah, thank you for connecting us with them.
Adolfo Brandes: Yeah, great that it worked. Perfect. So the next thing is something should have brought last meeting. We had a release retrospective or redwoods that ended a couple weeks ago at BTR and one of the many things that came up was How can we get more maintenance involvement as the sort of patch phrase right that can mean a lot of things and…
Feanil Patel: Yeah.
Adolfo Brandes: including not just maintenance right? So basically help fixing issues.
Feanil Patel: right
Adolfo Brandes: and I was tasked with bringing sort of a rough idea of what we're thinking here and I outlined that in the page and it's basically we want to have a single board that's going to capture all the issues including the testing issues We have a test plan for each release and the goal is for sumac to have an issue on that board for each of those tests. Right as well as issues that are caught outside the official testing. So anyway, the point is a single GitHub board with all of the issues of a release. And they will be marked.
Adolfo Brandes: When they're released blockers at least and maybe some other form of marking to clearly prioritize which ones BTR judges to be more important, and the hope is that maintenance during that release testing window will keep an eye on that and if they see something a critical issue that they think they could help with that they'd be free to start working on it right away without us having to ask each maintenance individually.
00:15:00
Adolfo Brandes: And of course that window will be announced probably here in this meeting as discussion forums Etc. So internally red finale we discussed how far we want to take. Getting maintenance involved and…
Feanil Patel: Yeah.
Adolfo Brandes: I think as a first tab, it's just going to be an informal like hey look, can you keep an eye on this?
Adolfo Brandes: And BTR will also of course try and tag maintenance whenever we feel. we could benefit from maintenance attention?
Feanil Patel: Yeah, yeah.
Adolfo Brandes: Also we are aware. There are cc's to help with this as well. And also PRS don't need to come just from CC's or maintenance. They could come from anybody in the community. It that it would be nice. For instance when there are repositories where nobody in BTR has any expertise in that we could talk to the About how to particular issue right even if the maintenance is not going to fix it.
Feanil Patel: Just the approach and…
Adolfo Brandes: a yeah. Yeah.
Feanil Patel: where to go. Yeah, that all makes perfect sense. the maintenance either should have that knowledge or should be willing to acquire that knowledge as the maintenance of a repository and help with debugging.
Adolfo Brandes: basically help a little with even…
Feanil Patel: Yeah. Yeah.
Adolfo Brandes: if not with creating PRS. particularly with important or blocking issues
Adolfo Brandes: Yeah, that's the real idea. Yeah.
Feanil Patel: That makes yeah. Yeah. are the issues that are created on this board going to be on the repos where they are relevant. So
Adolfo Brandes: Yeah.
Feanil Patel: because One of the things I think that tainers there's difficulty is that people are already watching too many boards, and I'm wondering if we can ask maintenance to Simply look for specific labels in the repos. They maintain or something. Simpler…
Adolfo Brandes: That's fair.
Feanil Patel: where they don't have to watch more boards, but they can just look out for things in the spaces where they're already supposed to be looking.
Adolfo Brandes: I think that's completely fair. So. Since of course issues from any repos can be on a board BTR can look at the board and then we move the issues to the once we find out where the issue actually is in terms of in the code. We'll be glad to move them to the Repository.
Feanil Patel: Yeah, that sounds great. And I think that will help us maintenance like involvement in the way that we want to moving forward. So I think that
Adolfo Brandes: Yeah, that's fair. that's an important. detail make sure to move the ticket to the repo so that maintenance don't have to be looking. at the board
Adolfo Brandes: Yeah, that's fair. I'll take this back to BTR. I don't think anybody's going to object.
Sarina Canelake: Clear that you're going to be able to move the issues because a lot of times the testers have literally no clue where the bug is coming from. They're like A product manager and they're clicking around and they're like, this button is broken.
Adolfo Brandes: Yeah, yeah. And that's likely be the bug tree ideas three others rule.
Sarina Canelake: so I think being just making it clear where everybody's roles lie. the ultimate end goal we want is these bugs are triage to the right repo, but we don't want to prevent people from reporting bugs if they don't know where the bug is from.
Feanil Patel: Absolutely. Yeah.
Adolfo Brandes: Yeah.
Kyle McCormick: Yeah. They might need to be a step where Treasures look up maintenance so that they can ask questions to figure out which repo the bug goes to. so I think this implies that maintenance should be watching their GitHub notifications like for pings
Feanil Patel: There and…
Kyle McCormick: otherwise they might
Feanil Patel: should be on the discussion forums for potential. Yeah, I guess it brings on the BTR issues you're saying Kyle?
00:20:00
Kyle McCormick: I guess yeah, unless this goes to ck. as much as I want to say everybody needs his own notifications for everything. another might be unrealistic so
Adolfo Brandes: And I mean the BTR roles somewhere in there somebody's going to try and reach out. the best of their ability right and…
Kyle McCormick: Yeah.
Adolfo Brandes: That includes trying to get some response somewhere, right?
Feanil Patel: Yeah, and I think that part of the process is probably going to be the most fuzzy because it involves individuals notification settings across systems. So as you guys try that out and…
Adolfo Brandes: Yeah, yeah.
Feanil Patel: you find what is and isn't working. Definitely let's loop back around and see we're as a maintainer you should already be getting notifications on the repos that you maintain if it makes sense for them to also get notifications when they are specifically mentioned in the BTR repo that is a thing that we should figure out if that's a requirement of being a maintainer and then update documentation and update expectations.
Adolfo Brandes: Yeah total I figure from other conversations that I maintenance should at least keep an eye on what's happening in the repo. issues PRS are opens if at some point in the future This is a baseline. I think that would be more than enough, But in the meantime, yeah, it's fuzzy will figure it out.
Feanil Patel: Yeah, yeah.
Adolfo Brandes: Yep, that's it.
Feanil Patel: Okay.
Adolfo Brandes: Thank you.
Feanil Patel: Yeah, thank you. Michelle
Michelle Philbrick: Yeah, I guess just looking for maybe some additional context Kristen from the two utnl team. On one of the PR's and fun and that course authoring she mentioned that the front end. Library content components one is going to merge into course offering. And I just wanted to double check on that because she asked me to leave some comments on the PRS because it looks like people might need to reopen PRS because they'll be closed when the repos merge. So I wasn't aware of it. So I just wanted to bring it to the group and see if anyone had any additional contacts on it before I start leaving comments.
Adolfo Brandes:
Adolfo Brandes: So the only context I have is that I had an inkling that this might happen.'s but I'll be glad to help coordinate the stuff.
Michelle Philbrick: Okay.
Adolfo Brandes: I just don't know. how many PRS we're talking about and which ones are more important than others.
Michelle Philbrick: Yeah, I don't.
Adolfo Brandes: quality glad to take a pass at it in case
Michelle Philbrick: Essentially the ask was can you add comments to the open PRS that they'll be closed when this is merged if they would like the work to still be merge. They'll need to open a PR in the course authoring repo.
Adolfo Brandes: What is that a sort of automated message at a portal PRS or? Are we?
Michelle Philbrick: no, I think it's just alerting people that have open PRS in the front end the components repo that repo is being merged and that they'll need to reopen it in the fun and the course authoring repo. Yeah.
Adolfo Brandes: I'm looking at it now. They're 11 yards.
Michelle Philbrick: Yeah, it's not a…
Adolfo Brandes: I think it's Yeah.
Michelle Philbrick: I just didn't know much about it. So before I went ahead and started doing that. I just wanted to make sure that everything was all set.
Feanil Patel:
Feanil Patel: I think an alternative proposal. Would be to review and merge or close those PRS before moving all of that content.
Adolfo Brandes: Yes.
Feanil Patel: Which would be a very reasonable thing to do I think.
Michelle Philbrick: so that's a touchy subject because a lot of those have been open for a very long time. And there's stop and…
Feanil Patel: no, I see that but also
Michelle Philbrick: the T&L had mentioned before that. they're maintaining it but I don't know if it was. a priority necessarily so
00:25:00
Michelle Philbrick: I think Jeremy had mentioned it was okay to give it to CC's if possible if things needed review so I don't yeah, I mean I can go back to Kristen with that and ask if that's possible, but Kyle
Kyle McCormick: Yeah, I suspect that some of these Po's are gonna be really old and I don't want to stop from making this change because it's a positive change. That said I find it a little weird that they're asking you to do this for them.
Michelle Philbrick: Yeah.
Kyle McCormick: They're the maintainer they…
Feanil Patel: Yeah, then they can do it.
Kyle McCormick: They have the best context they know You don't know when this marriage is gonna happen Michelle they do they can go to these PRS but a copy-based message in close 11prs.
Kyle McCormick: My recommendation is that you just kick that aspect of them.
Michelle Philbrick: Okay.
Sarina Canelake: Yeah, if you wanted to be helpful, you could provide some sample language. I mean I think one thing I keep finding is that it's so hard to close PRS people don't want to close PRS and I think there's a hesitancy to do so in part because they don't know what they're gonna say. It's just like in Adolfo was saying about maintenance being available for BTR issues. It's like what do maintainers say if they're not available? I mean, you need to say you need to say I am not free, right but We need people to actually sometimes say something a little bit negative or I can't help or I'm on vacation or whatever the case is in this case if they need to close PRS. They need to do it and it should come from them. But maybe you could say you need some help in figuring out the right words.
Michelle Philbrick: Okay.
Michelle Philbrick: Sounds good.
Robert Raposa: and Michelle, are you in contact with someone from Jeremy happens to be on vacation this week and Okay.
Michelle Philbrick: No, I was going to plan to just reach out to Kristen since she's the one that left the comment.
Robert Raposa: Yep, that sense.
Michelle Philbrick: It okay.
Robert Raposa: That sounds great. And I just want to let you know that Jeremy is on vacation. So if you need Jeremy's support,…
Michelle Philbrick: Okay. Thank you.
Robert Raposa: that'll be next week. Okay.
Feanil Patel: Yeah, that's great.
Feanil Patel: All And think Michelle you've got what you need.
Feanil Patel: The commerce deprecation that's this conversation right here. I wanted to follow up on this because there's been some back and forth on that thread but the upshot is basically there people who want to maintain the back ends, but no interest in the front ends. And also only as far as I can gather, it's multiple people but all will start working on one project for one organization. Given that information I'm leaning towards. Proceeding with the deprecation of commerce, but I wanted to sort of talk that through with other people.
Feanil Patel: they have been following along and have different opinions.
Sarina Canelake: I have an opinion. I mean, I think I raise this when I was trying to deprecate Cyprus and…
Feanil Patel: Is it?
Sarina Canelake: Max raced his hand and I really wanted to say way what he was offering with the benefits for the community because his organization uses these end tests and I wasn't getting a sense necessarily that other organizations were but the difference was he is committing to doing some amount of work. Or the community, right what a year on the call Max, I like to hear, I want to bring some of this to BTR. I think we can really amplify the use of this for the community and when We're going to deprecate a component and just one organization raises their hand and says this is useful to me and I want to develop it for my organization. I think that is the case where we have said in previous situations. We should continue with the deprecation and they can Fork it and continue to develop it in their own org.
Feanil Patel: Thanks for that.
00:30:00
Sarina Canelake: So given the information I haven't been following along but given the information you just relayed. That would be my inclination.
Feanil Patel: that's sort of where I'm leaning to so I appreciate the sound like the reassurance and agreement on that line of thinking because yeah It's Tricky when somebody wants to maintain pieces of a thing because I do want to support what the community needs but I think this is a situation where The uses are not I can't imagine a pathway where commerce sort of Turns around and becomes a thing that everybody in the community wants to use given sort of the complexity and level of technical debt that's living there.
Feanil Patel: All right in that case, I will write some replies and that'll be the other deprecation that I create master writing these down because now
Kyle McCormick: Speaking of repos and…
Feanil Patel: and that's
Kyle McCormick: writing a registered application right now. this will be a surprise to anyone who's involved in register.
Feanil Patel: okay.
Feanil Patel: Okay.
Feanil Patel: We are at 3933 and we are switching over to edx platform specific topics. It looks like so good timing everybody if you were here just for the Full maintenance working group and you don't have edx platform specific things to feel free to drop off. If you want to stick around for the platform sure talk about the X block work first. I think that's less ended. So maybe just a quick update there makes more sense. And then the resiliency topic I think is more ended.
Feanil Patel: Robert you want to ask a starting question or do you I just looking for an update.
Robert Raposa: No, I can ask my starting question which might sound more ended than you'd imagine which is more. Maybe this has been discussed already. It's the kind of work that. Seems like it could be. Potentially risky again and again for our deployments and I'm just wondering if we could have it could be here. It could be otherwise brainstorming ideas of how to minimize risk, because I know the risk is really
Feanil Patel: Yeah, yeah. It just happens to break you first,…
Robert Raposa: Our risk, but it's like all for the community,…
Feanil Patel: but it'll break everybody.
Robert Raposa: right?
Robert Raposa: Exactly, but It would have broken everybody. so yeah, so how can we make these kinds of changes and…
Feanil Patel: right
Robert Raposa: have it be minimized risk, and we've got in internal conversations around potential pipeline improvements and things like that to minimize some stuff but it would be great to Discuss that know if that's already been discussed and there's already an agreement of here's…
Kyle McCormick: Gonna take it right here for your Robert that I think you should check out.
Robert Raposa: how these will, more carefully be launched or whatever. So that's fine.
Kyle McCormick: That's basically a rollout plan.
Feanil Patel: Yeah, the experimentally enabled one as I think the one you want to take a look at.
Robert Raposa: and the man
Robert Raposa: Fairfax wants to restart before letting me know can you piss the link in…
Kyle McCormick: Sure thing. Yeah, in…
Kyle McCormick: I'll share my screen too.
Robert Raposa: Chat as well just for me.
Robert Raposa: Yeah. Thank you.
00:35:00
Robert Raposa: and should
Robert Raposa: just review this. Extract of fun. Sorry taking notes at the same time that I'm trying to talk.
Robert Raposa: yeah, maybe I'll just review and just see…
Kyle McCormick: Yeah.
Robert Raposa: if I have any thoughts of Possible improvements I feel like I read. That yeah,…
Feanil Patel: Yeah, yeah and for what it's worth we had a discussion about implementing this morning.
Robert Raposa: I don't know so I should get caught up on this again.
Feanil Patel: So it's very timely and a good time to sort of have more input about things you might want to change about this because the work hasn't started but it will start to
Kyle McCormick:
Robert Raposa: You got it. And does this feel I mean it so again, stop me if the answer is just read this first, but the questions with the toggles is everything set up for it's basically like
Robert Raposa: the code will be there and you don't have to worry about it and…
Feanil Patel: yeah.
Kyle McCormick: that
Robert Raposa: you'll just yield test on your own time with the toggles and then
Feanil Patel: Yeah.
Robert Raposa: And yeah, and I guess there's no other way to sort of. Other than tuggling it on. What we're using waffle so we could also do a slow rollout percentage rollout.
Kyle McCormick: you could yeah, I was gonna have them make course waffles just to give maximum flexibility to operators.
Feanil Patel: Yeah. Yeah, I think that's the way to go because I think doing a percentage roll out might be.
Robert Raposa: Yep.
Feanil Patel: complex because I don't think it doesn't have user Affinity, So a user might get it does.
Kyle McCormick: It does. Yeah, it's a welcome feature and…
Feanil Patel: Okay. that's
Kyle McCormick: we don't have to do anything on our side to allow them to use the percentage role that we just to find a course waffle and they can use the waffle interface that I want to do a stable or…
Feanil Patel: that's great.
Kyle McCormick: unstable percentage rollout.
Feanil Patel: Got it. Cool great.
Robert Raposa: And if you roll stuff back. it's all good and fun. I mean people can. Go from one to then back to the other and…
Kyle McCormick: Okay.
Robert Raposa: it's just…
Feanil Patel: Yeah.
Robert Raposa: what are they using Anyone particular moment or That sounds great. So I'll read more but that sounds like exactly…
Kyle McCormick: Yeah.
Robert Raposa: what I would want and…
Feanil Patel: that's to Jeff Kyle
Robert Raposa: wish for so
Kyle McCormick: It's easiest for us because I want us to be able to merge code in without.
Robert Raposa: nice
Kyle McCormick: Bring about it taking down your side at 3 AM.
Feanil Patel: Yeah.
Robert Raposa: exactly Long Live toggles All right. Thanks.
Kyle McCormick: Yes you
Feanil Patel: All right, and then resiliency generally we wanted to talk about.
Robert Raposa: Yeah, and I can't and
Feanil Patel: I got you.
Robert Raposa: why is my screen?
Robert Raposa: There we go.
Robert Raposa: And do we have the old? text anywhere so we had I think one was just an inform.
Feanil Patel: Yeah.
Robert Raposa: Which I can't remember Kyle. Were you here already for that discussion?
Feanil Patel: Yeah.
Robert Raposa: I think you are when just have Tim's changes.
Kyle McCormick: Is this the error page recursion?
Robert Raposa: so that's Yeah, yeah.
Robert Raposa: and then I think that was just a lead into your roadmap ticket and that new section that I added and you're like, yeah,…
Kyle McCormick: Okay.
Robert Raposa: that's definitely talk more and we just haven't had time to talk more so Do you?
00:40:00
Feanil Patel: One I think related to this that I'm curious about and I don't know policy of specifics you want to chat about is building these feedback loops for people who have apms and other tooling that will reveal these issues.
Feanil Patel: But was that other stuff you wanted to talk about?
Feanil Patel: Kyle
Robert Raposa: Are you asking America Yes.
Kyle McCormick: No, I don't have any ideas yet.
Kyle McCormick: Is the problem statement we need to build feedback loops for people with apms?
Robert Raposa: You got The other thing that go ahead. Sorry guys.
Feanil Patel: Yes, but let's step back for a second. But I think there was this high level architectural goal of making the platform more resilient. are there architectural changes that we could make to achieve that or is that just having good feedback loops for when we find resiliency issues in the wild? because the second thing feels like a thing whether or not it's good already exists, which is that we have feedback loops from Tim and you guys and other operators where they can report these issues and we would as maintenance prioritize them and figure out what we need to do about them except pull requests for improvements that they make and the other of Is there structural changes we want to make resiliency better?
Feanil Patel: Is I think a much bigger question that maybe is more relevant to us to talk about further.
Robert Raposa: Yeah, I think on the first one. just to touch on that for a moment is I had also introduced simply because As PR's or issues come up that are like, it just didn't feel like it fit under any of the other categories. But if…
Feanil Patel: Yeah. yeah.
Robert Raposa: if one of you felt like it fits on the other categories, or maybe you felt like This shouldn't be an encompassing ticket of all of our edx platform stuff. It's more like just to your second Point architectural things that we want to think about. And if it's just an individual ticket that is useful. We have some other place to track it or…
Feanil Patel: I think that's right.
Robert Raposa: some just miscellaneous.
Kyle McCormick: Yeah, it's that.
Robert Raposa: category or…
Robert Raposa: something like that
Kyle McCormick: It's for architectural work.
Kyle McCormick: I don't think we're gonna represent every issue. on this road map
Robert Raposa: Yeah, so an example ticket. that come up is throttling and sorry rate limiting has come up again and again as we're Adding it one by one when issues happen because we don't actually have it by default. we're also in the process of switching a whole bunch of apis over to drf, but we still don't have default rate limiting I think in the platform and so is that something that feels still like an individual ticket? which it is from one standpoint or does that feel more architectural to you or is architectural now that's really, how do we want to break things up or
Robert Raposa:
Feanil Patel: Yeah, I mean I…
Kyle McCormick: Yeah.
Feanil Patel: your heart
Kyle McCormick: this roadmap become more challenging that I originally thought. because at first it started as I was like It seems like there is this. Consensus that already existed for a lot of changes. We wanted to make to the platform and I'm like we better get this written down somewhere because it just exists in all these experience Engineers head. So let's put it somewhere and I wrote it down.
Kyle McCormick: the feedback I got from Ed, which I think is pretty Salient is
00:45:00
Feanil Patel: Be suppressed I have an opinion on this which is that I don't think we should do that. I think that it will be easy to list the problems and easy maybe not even easy to prioritize but let's imagine for the sake of argument that it's also easy to prioritize which problems which will work on. I think the step three which is having people actually work on it is the hardest part. and so one of the shifts that I've been making sort of internally to me and is in terms of making it clear that there are lots of important things and then allowing the community to sort of pick up the things that are important to them and sort of prioritizing basically one or two things at a time that are important to me that I'm working on instead of saying and the things are important to me or things that are like
Feanil Patel: with discussion with the rest of the accent team and with the community of what is the most important thing and not worry about what number two through because I think that can change quickly and also different organizations have different priorities. So trying to have a Centralized priority list doesn't work. It's much more useful to say Are you willing to work on any of this? Because the willingness to work on things is really the hard part.
Feanil Patel: So for example, awais has been leading the charge on this rest API conversion. Is it the most important thing that he could do with his time? Maybe not but also by completing it he unblocks our ability to have. Drf default rate limits which gives us better resiliency and is reducing and it also Auto documents all these apis that were previously not documented as rest apis in our edx platform docs and there are these improvements that happen as a result of it and…
Robert Raposa: Okay.
Feanil Patel: which is why it's on the list of architectural improvements in the first place. But if we were to be like should you work on this or dropping dropping Mongo capabilities is super valuable, but also,
Feanil Patel: Not necessarily easy or obvious for people to work on. so I'm a much bigger proponent of let get people to work on anything as opposed to choosing. What is the one right thing and documenting what the direction is is the sort of value we can provide as maintenance.
Robert Raposa: that's
Kyle McCormick: Yeah that last thing you said, I think maybe that's what I was trying to do. with document Direction…
Feanil Patel: day
Kyle McCormick: because I did not imagine this ticket to be an epic of epics that we work through sequentially or something because it's like everybody has their own priorities and…
Feanil Patel: right
Kyle McCormick: it was hoping to encourage people to Implement their priorities in a way that works towards this road that versus against it so perhaps an ADR or…
Feanil Patel: right
Kyle McCormick: of Porch the living document of this is our edx platform Arc vision. Here are some ideas. We have to work towards that but not actually. Track it or try to keep it up to date. Maybe that would be
Feanil Patel: I think it's something we revisit regularly and by regularly,…
Kyle McCormick: yeah.
Feanil Patel: I mean maybe twice a year but not something that we dedicate weekly time to keeping fresh and…
Robert Raposa: ation
Feanil Patel: up to date because I think that The overall and…
Robert Raposa: ation ation
Feanil Patel: and so I think that that may be a good litmus test for something that belongs on here or not. Which is this a fix that happens on the order of days or weeks? It's probably too small. Honestly for this is something that happens incrementally over a long period of time and people need to know which direction to move in and whenever they touch things they should be able to see what we want to do with it. Yeah, that probably belongs here.
00:50:00
Robert Raposa: Yeah, some thoughts. I like that this also just brings visibility to some subset of issues and makes it a little bit more digestible for someone who is
Robert Raposa: looking to help out in one of these areas like Hey,…
Feanil Patel: Okay. Yeah.
Robert Raposa: you know what? My problem is that. this is difficult to configure or whatever it is and I don't necessarily know exactly where I want to start with that and it's like here's some things that you could look at and…
Feanil Patel: right
Robert Raposa: which means it doesn't have to be a perfect list at all and it doesn't have to be wildly up to date but I
Robert Raposa: I like that this exists.
Feanil Patel: Yeah.
Robert Raposa: In terms of the earlier topics. I don't know if there's any categories that have come up where we don't think that that's A high enough priority at least to categorize here.
Feanil Patel: right
Robert Raposa: If that does come up we could also have a list Of categories of things that were not discussing here for now, and that could change if people come up and are like, what that's super important and…
Feanil Patel: Okay.
Robert Raposa: and I want to work on it, it's like, okay.
Robert Raposa: yeah, this tickets, and maybe it's not helpful or our conversations every once in a while of so On my backlog. I will probably keep things around like the drf. missing rate limiting thing especially because to introduce that kind of work would It's going to require to you to probably help they introduce it and…
Feanil Patel: Yeah.
Robert Raposa: make sure we're not breaking things and make sure we can fix it and give something reasonable.
Robert Raposa: That there's just a question of is it helpful to you all to have that? Highlighted someplace for discussions that we're thinking about and…
Feanil Patel: Yeah.
Robert Raposa: is this the ticket for that or is it just a separate conversation of hey, if you want to know edx platform tickets that are on our project board. cheers. Here's the link to see that.
Feanil Patel: I mean, I think one of the values this ticket provides is that not everything here is a ticket. and there can be ideas that are like, if we look at them like make module store go away, We're working on some of the things for that. but we know we want to make modules for go away in its current incarnation in the long run and there are lots of things that we could do towards that I think similarly for resiliency I think is a great topic in that sense of we know we want to improve and for I would even say resiliency and security kind of go hand in hand and we want to sort of improve the overall stability of the platform and their improvements we can make to do that that we don't have the resources to make all at once but that we don't want to lose sight of if you left tomorrow
Feanil Patel: I would want to capture rate limiting here so that we don't forget about it. Just because you're gone and that's what a lot of this stuff is. It's not they like we can do this tomorrow with it with a click of a button, but it's like Half of this would have been in Dave's head if we hadn't written it down and the other half would have been in Braden's head if we hadn't written it down and a bunch of it would have been in Kyle's head if we hadn't written it down and getting it all in one place. Where it's written down, even if it becomes out of date is helpful because it gives us context on architectural Vision that experienced engineers in this system have had over time.
Feanil Patel: it's a blend between a wish list and a Direction.
Kyle McCormick: yeah, something like
Feanil Patel: So I would say you don't even have to make a big ticket. You could say we don't have default rate limiting for apis and that's platform and that could be a sentence under make it more resilient or something more action oriented of edx platform should have a overall rate limit for apis and you don't have to answer how to do it or when to do it, but then it's here and that vision is clear.
Robert Raposa: Right and I think that issue does exist. So then it was a question of is that architectural and I think now we're arguing more for that this is just a useful place to track things that were in our brains and…
00:55:00
Feanil Patel: Yeah, put it on there.
Robert Raposa: they will be here so that…
Feanil Patel: Yeah.
Robert Raposa: if the brains change someone can come back to this be like this is useful.
Feanil Patel: Yeah. Exactly. Yeah.
Kyle McCormick: yeah, it's a brainstorming slash shared notepad for us settings platform maintenance and…
Feanil Patel: Yeah, exactly.
Kyle McCormick: I think the limits test for does it go here Would it be helpful to have it here and…
Robert Raposa: Yeah.
Kyle McCormick: it's so easy to edit that we can. If we disagree that's something shouldn't be that I'd say Aaron putting things there and…
Feanil Patel: Yeah.
Kyle McCormick: we can tidy reorganize as we go.
Robert Raposa: Yeah.
Feanil Patel: I'm like, things might close and we may want to remove them from this list because it's not valuable to track the things we've done here, but just to track what's not done here.
Kyle McCormick: Yeah, I think the thing I'm purposely not including on this is things that are facing features Except for when those user facing features directly improve the platforms architecture or stability. And so why Content Library is being the prime example of that
Feanil Patel: right
Kyle McCormick: so we don't need a ticket here for I don't know a new course mode unless that involves refactoring course notes.
Robert Raposa: right
Feanil Patel: This is mostly. The items where you are changing, the table legs out from underneath the table without anybody noticing. But now the table is a lot more stable and can hold more weight.
Kyle McCormick: Yeah, but we're deprecating one of those legs. So
Feanil Patel: You only need three. If they're in the right spots.
Kyle McCormick: there you go. Exactly.
Robert Raposa: And I see all these people on the unicycles and stuff like that, but I guess they're moving.
Feanil Patel: We're moving. maybe we only need to.
Robert Raposa: that exactly or even one.
Feanil Patel: we could try to move towards a wheelbarrow architecture of having one wheel and two legs.
Robert Raposa: exactly
Kyle McCormick: That's just a ad. Five years we're gonna kick ourselves if we miss that.
Feanil Patel: Yeah.
Robert Raposa: on it the topic of authentication so this is very quick because I don't have much to share other than I still know that's a thing that you is going to want to be dealing with and…
Feanil Patel: that
Robert Raposa: But I don't actually know what's going to happen with that. all
Feanil Patel: yeah, that makes sense, especially with all the Red Ventures shenanigans.
Robert Raposa: exactly
Feanil Patel: Yeah, do you think it's worth it to have a top level area for that stuff in and we already have a bunch of tickets in that space. I feel like the thing we're lacking there is North Star
Robert Raposa: yeah.
Feanil Patel: we have a lot of improvements we can make but not a Direction Where We want to end up.
Robert Raposa: I
Robert Raposa: I think that's the thing that's tied with the topic of resiliency is right now the identity service is the attic platform, right? and…
Feanil Patel: right
Robert Raposa: and to you. we may be moving towards that or we might be moving away from that and just allowing some external.
Feanil Patel: Yeah.
Robert Raposa: So to allow some external it would be how do you really allow for that? And if we're moving towards that X Platform One? How could we? Maybe make it a separate service or what or are we stuck with it being in the service and we just really don't want this service to go down, for other random stupid reasons, like someone's just calling some random endpoint harder than they should be. Yeah, so
Feanil Patel: Yeah.
Feanil Patel: I mean a thing
Feanil Patel: right and one of the things that I think we really do need from a Perspective it's just like to agree that we aren't enable skin. And skin is a standard for user management for third party services using a well defined set of apis. And the idea there, is that if you want to manage users across and it operator of many systems. If you want to manage identity across all of those systems skim is one of the known standards for doing that and if we supported that then people could more easily integrate us into their existing identity services. So Universe we have this already a little bit with oauth, but you pretty oauth is users get created at
01:00:00
Feanil Patel: Login time essentially, right? They go through an oath workflow with the identity provider and in our current system when they do that, we create a local user that maps to that remote user.
Feanil Patel: Skim would allow us to have apis where that remote system could Free provision and set up users in our that map to their other users so that
Feanil Patel: They don't have to necessarily share systems to have user management available
Robert Raposa: Yeah, it's not something and aware of so I don't know if that's the kind of thing where I don't know about an office zero. Type situation…
Feanil Patel: right
Robert Raposa: if we were to do something like that and would it be where skim could be Bill creating the user but not right when they log in or not when they come across but as they log in for the first time if they're not in the system so I don't know if it's applicable or not applicable to our potential use case.
Feanil Patel: Yeah. Yeah. I mean I kind of Imagine. That we honestly have enough. Auto enrollment capabilities right now that if I were Somebody like to you and I was like, I want to have authentication managed externally. I would just turn off being able to create users in. Edx platform I disable it introducing either the relevant Flags or removing the relevant Pages for myself. and then I would say there's a big login to you button, which is essentially an oauth flow with OCTA or whatever third-party system you want to use and it's just like Is hidden away from the user? the maintenance downshot of all of that is that there is always going to be and I think this is actually true for all of our services which is that there's always a local user object.
Feanil Patel: and I don't think we'll be able to get away from that because of the shortcomings of our
Feanil Patel: Shot set up that we have right now. Because the alternative would be to encode all of those permissions and capabilities into the job. And if you think about the level of permissioning we have around Can you access a course what your role is in that course what your role is in an org what your role is on the site all to track every course that you have access to and the role you have in it that jot would get too big and this is already a problem where we're chose not to run into. So I think you're gonna need a local representation of a user in our services no matter what.
Robert Raposa: language
Feanil Patel: And some way if we introduce apis to be able to manage user roles from an external system. That might be good enough.
Kyle McCormick: You guys aware of the roles and permissions project that XM is going to fund soon.
Robert Raposa: not exactly and I'm
Robert Raposa: I did know that no topic is a quick topic. So I brought up authentication and I'm like and now it's 10:05 and my apologies. So if we want to discuss that more maybe we should just put an action item for discussing rules and permissions or
Feanil Patel: yeah, I can do that. We'll do that for next time.
Robert Raposa: Yeah, and…
Kyle McCormick: yeah,…
Robert Raposa: and thank you Kyle for bringing up more stuff.
Kyle McCormick: I actually brought it up as a question because I don't know anything about it. I just heard about it and…
Robert Raposa: Okay.
Kyle McCormick: then said boy,…
Feanil Patel: it's really new. Yeah.
Kyle McCormick: cool. If we can.
Robert Raposa: I see but you did say Axiom, right? Okay.
Kyle McCormick: I did. Yeah.
Robert Raposa: Got it. I was like you didn't say to you so you're not looking for me to answer that question.
Feanil Patel: Moana no, this is really recent.
Kyle McCormick: Yeah, I think we've left the phase of being company where everybody knows everything and now we're yeah.
Robert Raposa: It got it. Sounds good.
Feanil Patel: Yeah, this is pretty new. It's being led from the product side, which is why you haven't seen as much information yet. Because I think we're trying to figure out what the user we're working down from the user experience. Yeah.
01:05:00
Kyle McCormick: Right Robert.
Robert Raposa: You can say that's what I need to drop. All right. Thanks.
Feanil Patel: but yeah, let's we can. I'm also realizing I won't be here next week Kyle. So you're in charge.
Kyle McCormick: Gotcha, okay. Then we're just not going to talk about it.
Feanil Patel: You can check the week after.
Kyle McCormick: That's good. Alright, take it easy.
Feanil Patel: Bye
Meeting ended after 01:05:38 👋
This editable transcript was computer generated and might contain errors. People can also change the text after it was created.