2024-05-16 Meeting notes
- Feanil Patel
Date
May 16, 2024
Participants
@Feanil Patel
@Jeremy Ristau
@Kyle McCormick
@Chintan Joshi
Previous TODOs
Discussion topics
Item | Presenter | Notes |
---|
Item | Presenter | Notes |
---|---|---|
Dropping Python 3.8/3.11 | Feanil |
|
wg-maintainers-edx-platform team | Kyle | https://github.com/openedx/edx-platform/pull/34746
|
Switching to GH Runners for edx-platform | Feanil |
|
Open edx-platform DEPRs (time permitting) |
| |
Prioritization for What Lands on the Main Branches |
|
|
Footer behind a Plugin Slot |
| For MFEs in tutor, we’re trying a new strategy for how the plugin slot would work.
|
Upstream Dockerfiles | Kyle |
|
|
|
|
Action items
Recording and Transcripts
edx-platform Maintenance Sub-Group (2024-05-16 09:06 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
Transcript
This editable transcript was computer generated and might contain errors. People can also change the text after it was created.
Feanil Patel: That looks more, correct. Hi, welcome everybody to the meeting now. We're recording and transcribing.
Feanil Patel: I will share my screen. So everybody is literally on the same page.
Feanil Patel: Thinking about it.
Feanil Patel: I don't know why it makes it presented to yourself also, because then I just have two copies of it instead of the faces. I want to see. Anyway, yeah, so a couple of different things. I think most of these to Do's are not related to Edx platform so I'm just going to skip over them for now. We'll come back to them in the next meeting.
Robert Raposa: zooming in just one or two clicks
Feanil Patel: Yeah. Got you.
Robert Raposa: Yeah, yeah on the dark side.
Feanil Patel: Yeah, is that better?
Robert Raposa: Okay. Yeah,…
Feanil Patel: Yeah. …
Robert Raposa: thank you.
Feanil Patel: you can go bigger. There's all this white space It's not super useful to us.
Robert Raposa: Yeah. Yeah, that's helpful. Thank you.
Feanil Patel: Moving back. That's okay. did you do so? Real quick yet. Sorry about the three eight dropping. I thought that with the other corporate coordination, we've been doing you guys had that on your plate already, but apologies for not confirming that.
Feanil Patel: it sounds like we'll get the three or three Eleven will ship
Feanil Patel: next week sounds like Wednesday.
Feanil Patel: Or is that a buy before that kind of a date?
Robert Raposa: I don't remember during who is the team that's doing that. All right.
Jeremy Ristau: Harvey Palm, yeah, their target is Wednesday 3 am essentially just the day after Mongo 7, so Tuesday…
Feanil Patel: Got it. Okay. Thanks.
Jeremy Ristau: if it very user and slips or whatever then it would be the day after that. That's essentially the
Feanil Patel: Can you make sure that drop it in WG maintenance when that's done or Ping On the pr?
Robert Raposa: Is it a PR that they'll merge and…
Feanil Patel: So that we know.
Robert Raposa: and will it have deployed to or is it not like that?
Feanil Patel: no, no, so it's a PR that I'll need to rebase and clean up because I've been dropping pieces of it that can go on and requirements of updated and…
Robert Raposa: Okay.
Feanil Patel: things like that, but it's the people.
Robert Raposa: Go.
Feanil Patel: Yeah, I think it's this PR
Robert Raposa: That's good.
Feanil Patel: In just so that way I know to pick that back up and ship it. And I'm gonna hold off on that. There's a related dot GitHub to update the defaults to 311 for all of the workflows that are reusable. that I'm gonna send out an announcement for that on discourse today, but that'll definitely affect you guys because for any repos that have not gone to 311. You may need to override a setting for the upgrade python requirements job.
Feanil Patel: I'll have the details in that post.
Robert Raposa: And is that just side? No Jeremy is that additional follow-up that RV bomb will need to take on and we'll take on and
Jeremy Ristau: Yeah, I mean right now they're pushing all internal repos to 311. So I think it's more of a race condition at this point.
Jeremy Ristau: Either they would have to do something or they're already doing the thing that would mean they don't have to do something depends on.
Feanil Patel: bread they can decide.
Robert Raposa: That okay.
Jeremy Ristau: their writing PRS and putting them to teams to review and merge. So it's not like an RV bomb controlled thing. it's all yeah.
00:05:00
Feanil Patel: and right Yeah, and that's what I'll put something in maintenance announce and that way you guys can cross like that to any internal teams that need that information as well.
Feanil Patel: But yeah, I probably won't merge that till after next Wednesday or whenever the python 311x platform that goes up.
Jeremy Ristau: Okay, I doubt all of our results will be on 311 by then. So.
Feanil Patel: and I think most
Feanil Patel: yeah, I think some people will definitely have to do a small update to the GitHub workflow to Adam one line change to run with three instead of three 11, but I will have a example of that in my post. So I think it should be a really easy change for people to pick up and make
Feanil Patel: all right. and then
Feanil Patel: Let's talk about WG maintainers at its platform Kyle.
Kyle McCormick: just to start I think maybe the way we presented this change was a little not the right way to start the conversation. But since here we are. I guess I want to express that you folks do show up to these meetings and you definitely do help with that X platform maintenance and I didn't want to give that impression by saying that.
Kyle McCormick: But by last meeting what we said in this so my bad if we did give that impression. you folks do a lot of people until you do a lot of people not to you. we aren't even talking about being on this team do a lot for edx platform. so what I'd like A set of people that I think should be in this team. Are people who are leading parts of the edx platform maintenance strategy
Jeremy Ristau:
Kyle McCormick: So it sounds like you guys want to be on this team. So what I'd ask you is what major parts would you like to take point on?
Jeremy Ristau: What's the strategy?
Jeremy Ristau: Other than maintain and…
Kyle McCormick: but
Jeremy Ristau: upgrade stuff.
Feanil Patel: so I think It's probably worth writing some of this down. But I think there's a lot of backlog of Maintenance that needs to be done to sort of get us into sort of a healthier State. I know that there is a lot of deprecations that need to be completed part of
Feanil Patel: I don't know if you guys have seen this sort of this is a more brainstorming than strategy, but I think it's like a pretty coherent. I'm gonna post it there for you.
Feanil Patel: these are some of the sort of high level things that At least Kyle and…
Jeremy Ristau: years
Feanil Patel: I have been talking about.
Feanil Patel: and I'll post this And the dock as well. Thank you.
Robert Raposa: because I mean One sign in to go through the stock which I will and so one interesting. thing that you're implying Kyle and thing that I'm wondering about is the difference between
Robert Raposa: to be in this team You're saying means to be like an individual that takes one of these things and is signing up to be the leader of that it's one thing that you're proposing where I think Jeremy and I were thinking more like this. Team right here that's meeting that is leading the edx platform related issues and trying to figure out how to do it is that level of hey? here's the topic that we need to discuss and we need to move forward and then maybe there's action items that come out of that, but that it's not like some separate thing, but maybe Jeremy has different set of thoughts on us, but it's
Robert Raposa: Yeah, so I was thinking more had these meetings and this team, potentially content and whoever is like the team we're going to discuss these topics and push them forward and figure out how that needs to happen as a team.
00:10:00
Robert Raposa: that's one way in which you get to be on this team and continue to push forward with what needs to happen first the maintenance of X platform.
Feanil Patel: Yeah, I think. I'm going to project a little bit and apologize. But here's the thing that I think an underlying fear is that while you is going to be helping to do a lot of the work the coordinating and the sort of high level. Thinking about this stuff will not be done it to you. And I think that the thing that maybe Kyle you're worried about is that accurate I think that. What might be more true is that in this moment?
Feanil Patel: Maybe this is what this is me. I think one of my fears is that you guys are gonna be helping with a lot of work but not leading a lot of work. I think that's true today and maybe that's not true in the future and if that's the case and it's like a situation where it's like hey right now while we're in the year of our life to you where we need to figure a bunch of s*** out. We won't be able to help with it, but that we want to be participating and we want to Future in the future be leading some of these efforts to make a true platform out of edx platform. that's a very reason like that would be a reasonable response. I think.
Robert Raposa: Yeah, I guess part of it means getting into the details of choosing a particular thing and what it means to you to be leading a thing and how that requires either an individual or a organization that's different than this group. Doing whatever that leading is.
Feanil Patel: Yeah, leading in my mind. I think it requires all of the other follow-up work that the python 311 upgrade required or that the paperwork has required as examples right like making sure that You are communicating the changes out making sure that the Right tickets are being created making sure that people are taking on the work and that if the work is stagnating you're doing it nothing in some degree. RB bomb is doing a bunch of the work, but I think Jeremy and you are kind of leading it and I just want to kind of figure out what that role looks like over there.
Jeremy Ristau: I mean my perspective project managers are not leader That's a different role. We're talking about putting people in a leadership Not a project management role so saying …
Feanil Patel: Yeah. Yeah.
Jeremy Ristau: because we're the project manager we deserve to be the leader role and no one else does is a little iffy you want to be like I just want you to make sure you tease apart those two roles when you're talking like I'm sitting in both chairs now,…
Feanil Patel: Yeah. Yeah.
Jeremy Ristau: so therefore I be careful when
Feanil Patel: No, no that's totally fair. I think it's more that A lot of the work tends to be in the decision making in the midst of project management and I think having those two be separate roles is detrimental for the size of work that we're trying to do here because there's more back and forth that needs to happen. I think having a person who can do both of those things be a sort of top level maintainer is really valuable for something as big as X platform because the number of micro decisions that had to get made while in the midst of a ticket that would have taken four days of back and forth if there was multiple people and was immediate I think is extremely valuable because of how quickly we want to move here and…
Jeremy Ristau: like Usama and…
Feanil Patel: and how big it is.
Jeremy Ristau: Subhan make decisions too Doula does as well so maybe Okay,…
Feanil Patel: Yeah, yeah.
Jeremy Ristau: I think the biggest concern that I have is that I got an entire team who's almost all time dedicated to maintenance activities now.
Feanil Patel: Yeah.
Feanil Patel: right
Jeremy Ristau: I need to know what they are going to spend their time also need to be able to say no and…
Feanil Patel: Yeah.
Jeremy Ristau: if I am not involved if no one you is a part of the decision-making for what goes when then we can't pull back.
Feanil Patel: Okay. Yeah, I
Jeremy Ristau: My concern I think Robert has a ton of other different concerns, but mine specifically is we're dedicating a large amount of hours of people to do work. if it's like Action would like to take the platform this way and…
Feanil Patel: Yeah.
Jeremy Ristau: it's gonna F over to you for a great period of time and they don't really care about the timing or maybe they care but it doesn't really matter because they've got some other external thing that they need to deliver for like that. It's kind of very difficult.
00:15:00
Feanil Patel: Yeah, yeah. I hear you.
Jeremy Ristau: and that I
Feanil Patel: You guys are feeling like there's a seat at the table that allows you to be able to.
Feanil Patel: To sort of represent that more strongly as a named maintenance to a person participating in the maintenance.
Jeremy Ristau: yeah, I mean that's aside from the previous conversations. We've had around shared ownership and shared maintenance. which become across
Feanil Patel: Yeah. Let me sense.
Feanil Patel: Yeah. All right.
Jeremy Ristau: but I Kyle's seemed not aligned with your perspective. So I'm interested to hear what and that
Feanil Patel: Yeah.
Kyle McCormick: Yeah.
Kyle McCormick: I guess going back a little showing up to this meeting. is leadership I don't think discussing things on its own as leadership. I think. Taking what we do here.
Kyle McCormick: Making sure things get done in edx platform is what I'm looking for.
Kyle McCormick: I don't know. That's all I got.
Feanil Patel: Yeah. Yeah, I think. And kind of hearing your perspective Jeremy. I think that that is very reasonable. And I think that maybe is a perspective that I was fully grocking or understanding so
Jeremy Ristau: And if it's George instead of me, I don't really care what I need to make.
Feanil Patel: Yeah, yeah.
Jeremy Ristau: Do you have forever? Because to you guys dedicated hours to it. and I mean this is outside of the other work that teams are doing aperture on three pose and
Jeremy Ristau: Like that other maintenance cost that we pay but there's a dedicated team of people that are managed to push maintenance forward. That team is funded to you.
Feanil Patel: right
Jeremy Ristau: So we need to have control over that. Otherwise, it's an axum team or it's a community.
Feanil Patel: Yeah, I want to make sure we have time for these other topics. So this is super useful discussion. Let me think more about it also, and maybe we can continue it. again,…
Jeremy Ristau: pacing
Feanil Patel: we're not a ration. But yeah, we consider as anchor catch up a little bit more next time as people's thoughts. Gel and Thank you everybody for your thoughts. I think we will come out of this in a very reasonable place but It's good to do the real quick update on the switching GitHub actions Runners from edx platform that work isn't done yet.
Jeremy Ristau: Yeah.
Feanil Patel: So unfortunately we'll need those workers running for a little bit longer but it is in progress. I had planned on picking it up, but then the 3/8 stuff took longer dropping the old stuff a little bit longer. So this was the thing I was going to pick up after that. I've asked somebody on the X improvements team to start taking a look at it so that they can get that prepped that is ready to go just behind dropping 311 or dropping three eight.
Jeremy Ristau: Yeah, I mean just let me know the timeline. It's 10K a month for us. So
Feanil Patel: Yeah, it should be like if it's not done in the next. week and a half. I'm gonna pick it back up myself and get it over the line. There's a lot of simplifications and…
Jeremy Ristau: Thank you.
Feanil Patel: enabling capabilities we get as soon as we land that in terms of maintainability of the edx platform, so
Jeremy Ristau: Thanks and Alex trebka and SRE also said if you need help, you can reach out to him, but I think all the access is all on accident side. So. They turn off the machines just gonna break stuff. So.
Feanil Patel: Yeah, yeah, it's super. Yeah, I have the 80% PR here. I just didn't have the time to go figure out what those failures were and what was going on with it. but somebody else is going to pick it up, so they'll hopefully have a little bit more capacity for it.
Jeremy Ristau: nice, thank
Feanil Patel: Yeah.
Feanil Patel: edx platform deprecations
Kyle McCormick: Yeah, I brought this up. I mean I think in general it'd be good. or In this meeting to look at the deprecations that are open on a repo.
00:20:00
Kyle McCormick: There's a particular one that I think I want to focus on. So if you click into the paver Decker.
Kyle McCormick: so I know I've seen a few Dippers recently. Some have bigger migration efforts than others. This one is actually a really small migration effort as far as I can tell from all There's a table down there that says hey, if you're using these asset commands specifically the ones that are say assets on the left replace them with these ones.
Kyle McCormick: you folks would need to do that in your configuration Fork And that's the entire thing. And I didn't be happy to review those to make sure they match with.
Kyle McCormick: Yeah, just to make sure they're correct. The benefit to the community with that done is that we can make image tools faster in images smaller by removing paper requirements from edx platforms-based requirements. so
Kyle McCormick: Yeah, I'm hoping to remove this soonish.
Kyle McCormick: Open to talk about timelines, but I think of all the debtors of open this is the one I'm really itching to get closed out.
Jeremy Ristau: is there a timeline on the pepper
Kyle McCormick: It's already accepted in five days.
Feanil Patel: In terms of when it's going to be done, but we're targeting a sumac.
Jeremy Ristau: yeah, but Targeting sumacs today or five months from now, right? So
Feanil Patel: So, I believe Kyle has merged all of the commands that are the migrate after commands now exist in the next platform today.
Kyle McCormick: Yeah, unless it says needs to implement which isn't the case for any of the asset commands.
Feanil Patel: and so I think that Kyle correct me if I'm wrong, but I think the critical thing is that you're not going to delete paver until after sumac is cut.
Kyle McCormick: paver is Sumac will not have paper.
Feanil Patel:
Kyle McCormick: I'd like to delete ever between now and sumac.
Feanil Patel: McCormick Got it. Okay, so
Jeremy Ristau: And so in your timeline is whenever people say they've done this.
Kyle McCormick: Basically whenever you guys say you've done it. The community is all set…
Jeremy Ristau: Yeah.
Kyle McCormick: because they're on the named releases and the new stuff is already in the interlace.
Jeremy Ristau: Right, of course.
Feanil Patel: To tutor. Yeah.
Feanil Patel: So I think the urgency goes up as we approach to Mac, but sometime between here and sumac this just needs to get sequenced on the TU side is what you're saying Kyle.
Kyle McCormick: Yeah, it is. Let's ask.
Robert Raposa: Okay.
Robert Raposa: so I mean just I'll continue to repeat the timing thing.
Feanil Patel: really good
Robert Raposa: So this doesn't seem like a big deal and it would be great. I'm with you you actually be able to land your thing and get the benefit that you want and I'll just repeat our timing thing that the Ended by end of June. We turn off New Relic and we'll be doing lots of more breathing. so this could be a potentially very simple small test that happens At the beginning of July, it could be a simple small test that happens sometime in June depending on where we get to.
Jeremy Ristau: is there also a stack rank of Deckers here or…
Robert Raposa: but
Jeremy Ristau: work so RV bomb could pick this up but
Jeremy Ristau: could go before work on python 312 or it could go before other things that are going on. How can I get a sense from axon or from the community about what you consider to be more important than other things for a particular release?
Kyle McCormick: great question
Feanil Patel: I wonder if a thing that we're missing is this specific to you prioritization for the work you guys have to do because you're on the oldest deployment and have a bunch of customization. So this and 312 in particular I would say this is before 312, but I don't actually Have a list of all of the things to you need to do. Specifically in front of me and I wondering if you guys are tracking that somewhere already or if that's a thing. We should start tracking somewhere.
00:25:00
Robert Raposa: I have some of that tract on our board. and will be adjusted into a different ticket, but I don't have all of it try. but
Jeremy Ristau: The size of effort is our problem the prioritization of Value is your problem that you need to tell.
Feanil Patel: Yeah, yeah.
Jeremy Ristau: Where to Kyle's go Where does the deprecation of insights where does all those things that you'd like to accomplish? What is the stack rank of those so that we can internally about this is too big for May or…
Feanil Patel: true
Jeremy Ristau: it would be very helpful to have that.
Feanil Patel: Okay. Yeah.
Kyle McCormick: Familiar risk of bleeding into the next meeting. I wonder if this is actually a project wide thing where we have these things that we want to land on main branches not necessarily named releases and there's an ordering Is a priority that we want them all to be done by sumac…
Jeremy Ristau: Yeah.
Feanil Patel: Yeah.
Kyle McCormick: but there's a priority ordering within that six month. land window and
Jeremy Ristau: Yeah. What…
Feanil Patel: friends
Jeremy Ristau: what would you be? Okay with not making it into sumac or that kind of thing. let's put those lower on the list. Absolutely. That'd be super helpful for me.
Feanil Patel: That sounds good. Yeah, let's maybe take a look at the maintenance board. Kyle maybe even I can take that async as a first one off. because that board needs more organization now that 311 312 is Behind us for sumac and…
Kyle McCormick: Yeah.
Feanil Patel: we can start adding. I think some of these things to that board a corresponding ticket or even the same ticket to the maintenance board and then the prioritization on the maintenance board is how we focus on that stuff.
Kyle McCormick: cool Yeah, this isn't just an X platform problem. I think it just particularly.
Feanil Patel: Yeah.
Kyle McCormick: happens the most in edx platform by Nature
Feanil Patel: Yeah, I can think of the plug-in slots versus feature Flags is another conversation that I know we've had a bunch in the recent months. And figuring out which plugin slots need to be implemented instead of overridden with forks is another version of this.
Kyle McCormick: Thanks for joining everyone. We're just kind of wrapping up. the last topic which started as an edx platform specific thing, but I think it's relevant for maintenance in general where we have these priorities that we want to learn to name releases, but even more granularly we have priorities for what we want to land before other On main branches which affects people who to play half of me particularly to you. So
Feanil Patel: yeah, I think we can do that. cool and welcome everybody to the maintenance working group meeting. Let's kick it off with the real quick review.
Feanil Patel: Michelle I think you've put in this accent label request that's done. Does that sound right?
Feanil Patel: Yeah, and then I think Serena hasn't gotten a chance to do this update of oep 10. Or up whichever that elap is with the config stuff, but that's pretty much so we've got some time.
Feanil Patel: Kyle you did this creating PR's. To reduce the number of testnetics platform already.
Kyle McCormick: Yep.
Feanil Patel: And then I further reduce them. We've dropped everything except for three 11 at this point, I believe. so that landed the day before yesterday. So if you have an old FedEx platform and you want it to run faster. Make sure you read base is off of Master.
Feanil Patel: and then some yeah.
Jeremy Ristau: That just for clarity. It was solely. dropping old versions Yeah, okay.
Feanil Patel: Yeah, yeah. Yeah, it was solely dropping python three eight testing and Mongo 4 testing. We are currently testing with only three 11 and Mongo seven. And my SQL 8 Which is less relevant, but still not. yeah, I think it's like
Robert Raposa: know Sorry, keep going. No and…
Feanil Patel: No, please God.
Robert Raposa: finish what your system what you're saying.
Feanil Patel: I think there's about 30 tests shards that are currently running now as a result. There's a bunch of smaller ones and about 16 or so unit test ones.
00:30:00
Robert Raposa: and all I was gonna say is we had a script somewhere that we've used for open PRS that may need to be rebased for some reason so that we could just put comments out and notify the people that it's relevant to of hey, you can give this a rebasing so I could see if I could uncover that and have that more widely useful.
Feanil Patel: Yeah. Yeah, that'd be great.
Feanil Patel: And then models of Maintenance stock these two. I'm going to start working on next week, I believe so, hopefully we'll have some more progress on that space. And I think Jeremy you're continuing to update ownership as you get time and space for that.
Jeremy Ristau: Yeah, I mean I would call that blocked by time and space.
Feanil Patel: Okay, cool. I'm gonna be trying to get the maintenance spreadsheet to be more up to date and figure out which things to Target next for looking for new maintenance. So as a part of that I'll be following up with all the people on who volunteered today on maintenance but have not yet managed to update their catalog info files. I may end up updating your repos as a result and just tagging you on some PRS if I do that so be on the lookout for that if I get the time for that.
Feanil Patel: Okay.
Feanil Patel: And then I believe that's everything on here is anything else anybody wants to cover? I know Kyle has a new Dipper that we want to talk through.
Feanil Patel: Okay.
Kyle McCormick: This one could go on for a little bit. So there's anything more specific anyone wants to bring up. And cut in front of me in line. Go ahead.
Feanil Patel: It I guess a real.
Adolfo Brandes: It's sort of …
Adolfo Brandes: very good. It's sort of heads up since we get more to you in this meeting and case. You're not aware. We're trying to put the footer behind the plugin slot in the mfgs that are officially released. Or rather the ones that are in tutor.
Feanil Patel: You go Adolfo.
Adolfo Brandes: And our first attempt had a negative impact to you build. So we reverted that and we're trying something else that we're hoping does not happen. In fact. And since we're looking to have this Redwood as the backboard. Sort of a heads up slash.
Adolfo Brandes: if something goes wrong will react with quickly, but we're looking to fix it quickly to kind of thing. So yeah, that's all.
Robert Raposa: Are you still working with someone directly that just helps you have a really quick feedback loop on that.
Adolfo Brandes: So It's just a reality of it is that there are many mfps and many different teams right responsible for each one.
Robert Raposa: very
Adolfo Brandes: Which is why I'm bringing it up here. So in some cases the loop is really quick such as with Kelly's team and the learner dashboard. and then some others it may even take a while or the problem to show up if there is any
Adolfo Brandes: though on the cases with the fast loop we tend to find out what the problem is and then fix it for everything else quickly as well. So
Adolfo Brandes: I don't know. It's just like but is there a better way to do it than this is…
Robert Raposa: Not that I'm just saying…
Adolfo Brandes: what I'm asking.
Robert Raposa: is it's a problem that we are going to find everywhere. Then. It sounds like working within one MFE where you have a quick Loop first sounds like the right thing to do get all the problems shaken out and then fix across.
Adolfo Brandes: Alright, cool.
Robert Raposa: Yeah.
Feanil Patel: The Lord Jeremy. I think you Robert cover my kitchen.
Jeremy Ristau: My question is around the fix it everywhere point. So my question is are we making the new plug-in slot work with all the different ways that the footer is implemented or are we aligning? All of the mfe's to implement the footer in the same way one of those feels like good maintenance and the other one feels like trying to shove it out the door. So I'm wondering from a maintenance like this. That's this group's topic.
Adolfo Brandes: So I guess our first attempt was more the latter and now we're more in the former. So where we're trying to make it implicit API of forking the footer And then installing it I guess we have to tubular in your case. And if you tutor it was via a manual npm install. That's still gonna work. and then tension is If you're not interested in using that as a plugin, you don't need to do anything for this really. so as backwards compatible as we can make I don't know…
00:35:00
Feanil Patel: the go right
Adolfo Brandes: if it answers your question anyway.
Feanil Patel: Yeah, I think it's half of it which is the other half of it is that so the goal is that we wanted to in Redwood because this is an attempt to backport to Redwood make it so that the things that are there currently will all work. and then the goal is in sumac right Adolfo, you're dropping the ability to do essentially what we're like we're dropping the ability to have the hacky flickers because we're gonna require a plugin slot in the defaults.
Adolfo Brandes: Yeah, or at least we won't explicitly support it because the way we're doing it now is We don't need to concern ourselves anymore in the future right if you want to Fork You'll be able to still probably but it's not gonna be in it's something we support. All right. Yeah, there's gonna be and
Jeremy Ristau: maybe like a parking lot or offline question would be If we're not explicitly removing it, but we're offering support that we're offering the ability to status quo but not like long-going support. Is there a reason to shove it into Redwood or should this just be a sumac thing?
Adolfo Brandes: the reason we wanted in Redwood is so we can Have a relatively simple. use space for a plugin offered widely sooner rather than later so that by sumac we have more information to make decisions on are we doing plugins right are people actually using it this work it doesn't because otherwise We have a few slots. but they're mostly in places where we are removing business specific logic, right and not from us, a product point of view of saying Let's offer slots to people so they can use it and the plugins the footer slot would be that makes
Feanil Patel: And I think it the goal is to get feedback loop on the plugin slot technology sooner rather than later so that we can have a better iteration of it by seeing that.
Adolfo Brandes: exact
Feanil Patel: Adolfo I put an action item for you and me to sort of right up a Deborah slash maintenance for the impact. when we go to the sort of final behavior that I think is going to impact many people and I want to make sure we have the comms for it because if you drop it, for example, the day after sumac has cut I imagine a bunch of to you repos will probably break and so they want to prioritize that fixed sooner rather than later.
Adolfo Brandes: Yeah.
Feanil Patel: For anybody in the community, they want to be able to know and try those things out in Redwood. Once your stuff is out, and not wait till sumac where it has to be fixed. So I think there's some maintenance last debber overlap communication that should happen around this stuff that Yeah.
Adolfo Brandes: definitely. No, the whole point of revisiting our implementation is to Have time to do that, right? So yeah.
Feanil Patel: Yeah, I put an item here. Just so like you and I can catch up and I can help write the Decker the maintenance to get for that.
Adolfo Brandes: All right, cool.
Feanil Patel: because yeah, For Jeremy and Robert, I think. The target is Sumac which means that's probably one of the things that needs to go on the prioritized list between now and sumac as far as you guys are concerned and that way, it's on there.
Jeremy Ristau: Right and just as a mental model to constantly keep in your head the more time that you want to give people of bacon time after you deploy something. We're on the opposite side of that. That means we have less time to be able to plan and react to it. So, …
Adolfo Brandes: That I got it.
Jeremy Ristau: it's Yeah.
Feanil Patel: Yeah.
Adolfo Brandes: So we're talking specifically about the footer, but it's probably going to include the header as soon as well.
Feanil Patel: All up together Yeah.
Adolfo Brandes: So we're probably going to Do a one to punch on that. Yeah.
Feanil Patel: Yeah, and the goal is I think they're like
Feanil Patel: I think this is the thing that we talked about it and we arrived that as we talked more about it, which is the idea of a more mature drop it even though we have a bunch of implicit apis for how we interact. I like different components in the system interact and as we move to explicit apis, I think rather than just dropping the implicit ones. We need to make sure that we create the old implicit apis as the current apis and deprecating communicate and either work around them to some degree as much as we can and communicate that until we are ready to have official apis. So I think that's A place where we're still maturing, but meant that sort of thought process is now being applied more widely.
00:40:00
Adolfo Brandes: Yeah, totally in agreement.
Feanil Patel: All take the last 13 minutes to talk about Docker files Kyle.
Kyle McCormick: They're speaking of implicit apis. There's a deprecation ticket here in communication. So just to remind everyone that means that it's not a decision yet. It's just a discussion that becomes a decision in a month if there's no
Kyle McCormick: no, big notes that come out of it, but the gist is that we have these Docker files and every Deployable application repository so edx platform credentials Etc I know that these Docker files. and are used for devsec.
Kyle McCormick: I know that to you deploy some of them to production. I don't know any other uses of them. And so as we're getting to so now that devsack is deprecated, and we're gradually cleaning up.
Kyle McCormick: the tendrils of dead stack I guess a few things. I'm wondering whether other folks in the community use these for production and we just don't know that I'm also wondering if there's value in having Docker files in the Upstream repositories. Even though just as a good thing. but if we do keep them on wondering how we maintain these in a way that is these are the community supported Docker files and not these are to use Docker files that we just happen to have in the Upstream repositories.
Kyle McCormick: What do folks think?
Robert Raposa: Sounds like very reasonable questions.
Robert Raposa: I don't actually know answers to them. I don't know. I'll stop.
Feanil Patel: governments
Kyle McCormick: What do you think? That's
Maksim Sokolskiy: Yeah. Yeah, I can say that. we kind of used this files in production but not in a way they not as is but some sort of
Maksim Sokolskiy: and no manual or instructions of how this service should be built. And actually let's hope so from my opinion. This files are not ready for production. We modify everything but we use it as a template but Maybe from a half a year ago. We started to use a tutorial local files as a template files or even you use it as this interaction. So yeah was answer is we kinda used? all these files but in some way
Maksim Sokolskiy: and from my opinion it should be some kind of A procedure how each service should be bootstrapped.
Kyle McCormick: Right like an operations guide. Here's how you set it up.
Maksim Sokolskiy: Yeah.
Maksim Sokolskiy: Only be a problem if we will lose a tutor tomorrow and now tutorial disappears and we'll lose everything.
Kyle McCormick: so if the code
Maksim Sokolskiy: it's imagine imaginable thing, but
Kyle McCormick: I mean, there's definitely question of contingency. Right. Now the open edx Community does rely a lot on Twitter.
Kyle McCormick: even if the project disappeared for somehow It's open source, so we could begin maintaining it as a community if we really needed to. I think there is value in saying. We should be able to deploy open edx people do deploy open edx, not just to other people deploy open edx without tutor and It's good thing to be able to support that but are the docker files part of that strategy.
00:45:00
Feanil Patel: I think might sort of personal gut feeling on this. Is that I like when a repo has a really simple Docker file that's sort of not super production grade, but that starts up in a product with a production G unicorn or you is here something and starts up like your mint to start in production and shows you here I can read one of those Docker files and I can say this needs python 311 and this needs these system packages and this needs these requirements installed and all of that is sort of as a piece of documentation, it's really valuable. but only because that's a lot easier to confirm working documentation then the readme for example.
Feanil Patel: So I like it as a piece of Festival dock.
Feanil Patel: But on the flip side, I also know that it takes a long time to build them right now. And as we're about to put more work on the standard GitHub Runners. I'm not excited to run them if they're not. useful for a long time
Feanil Patel: I think if you have other opinions or you have other people who you think will have opinions, please Point them to that deeper ticket and have them chime in sometime over the next month.
Feanil Patel: Because I think unless there are sort of. Big reasons why it's worth keeping them. The current decision on that ticket is to drop them. Is that right Kyle?
Kyle McCormick: That is the proposed decision. Yeah.
Feanil Patel: Yeah.
Jeremy Ristau: And sorry, did you learn anything here? That will change that?
Kyle McCormick: I think I heard a couple reasons. In favor of keeping them, but not enough to sway it right now. So I'd like personally before I changed my mind, I'd like to hear more votes in favor.
Jeremy Ristau: Okay, but just to you count. Okay, because it sounded like you're saying if it's too specific it doesn't matter. So does to you actually have an impact on this decision?
Kyle McCormick: I factor in what I think you will say, which is that. we use these now and removing them further from the repositories would be disruptive. Is there more? That I should have in mind.
Jeremy Ristau: nope, just like I'm wondering if that plus these things changes the proposal or did you collect feedback to hear feedback or did you collect feedback to change your proposal? I don't know what the next step is here.
Kyle McCormick: The next step. Is that over the month between now and the acceptance say if more people Come out in favor of keeping them.
Kyle McCormick: I'd be open to changing my mind.
Feanil Patel: I think Yeah.
Kyle McCormick: Just from this 10 minutes that hasn't happened yet.
Feanil Patel: yeah, and I think there's the acceptance and then there's sort of sequencing it in turn around other maintenance and getting it actually landed and
Jeremy Ristau: yeah, yeah the previous conversation on stack ranking, but I think I heard Max say there's some value in it and…
Feanil Patel: yeah.
Jeremy Ristau: finale saying there's some value in it and two you saying in some value for us in it?
Robert Raposa: And I didn't actually look yet. So I don't know what the proposal is in terms of timeline. But again, I feel like for Community related things timeline has typically been let's give you at least a full release cycle to even think about this and when it comes to we think to you as the only thing that's actually affected then it's like let's do this as soon as possible. So we need to just figure out what the timing is. And keep that in mind having a whole. six months to think about it and prepare it's very different than let's come and get it in next month or the following week or They're two very different kinds of discussions.
00:50:00
Kyle McCormick: Yes for what? It's worth. This doctor file does use the dead stack? settings file
Kyle McCormick: which is accepted for deprecation and removal. so
Kyle McCormick: I guess there's some overlap between the devsack Dapper and…
Feanil Patel: this
Kyle McCormick: the docker file Dapper, but I'm trying to separate them because They do seem like two separate. Things that would be removed.
Feanil Patel: yeah, this one I think is not being used in production. But I know some of the other services this one is just for devsec.
Kyle McCormick: Yeah.
Kyle McCormick: So I think we did talk about reasons to keep them and I hear the reason that those by themselves are not convincing me is because there's a cost to keeping them. like finial said it takes up Community Action Runner time. We have Docker Hub. We
Kyle McCormick: we haven't X and managed Docker Hub that hosts these images with the open edx brand on them that we don't look at it all we don't Monitor and there's definitely I don't want to be the source of a supply chain issue where people are depending on these images that we don't think about.
Kyle McCormick: And also whenever there's an upgrade. People there's this implicit like hey, you got to update the docker file, even though the community at large isn't using them. So it's the same set of arguments that you have for anything that Isn't being used or supported by the community.
Kyle McCormick: if we think it's worth maintaining these that's something we can decide. I would want to maintain them more. Intentionally, then we are right now as a community.
Feanil Patel: yeah, anything like
Kyle McCormick: I want them to be in the Upgrade run books that I want. Xm2 pay some attention to dockerhub and the images that are being uploaded there to make sure that we're not Distributing things that could lead to a security vulnerability for someone who says, look open Notes images. Let's use these
Feanil Patel: Yeah, we're running up against time. So definitely put more thoughts on that Dipper. I think it's sort of make a note on what I said, I think for them to be useful for what I want. They might not be useful for the two you production deployment anymore. And so there are some sort of conflicts there about a Docker file might be useful the ones that I see here might not be useful because I think they're especially the course Discovery one being a good example.
Feanil Patel: It's like running a production kubernetes settings file that I think expects it to be run a very specific way.
Robert Raposa: And it sounds reasonable that you'd want to separate those two things to have documentation Related Dr. File not be like,…
Feanil Patel: Yeah, yeah.
Robert Raposa: can we update our docs without breaking to you production?
Feanil Patel: right Yeah.
Robert Raposa: So that's reasonable. so
Feanil Patel: and thank you everyone. We'll see you next time.
Maksim Sokolskiy: people
Meeting ended after 00:54:39 👋