...
\uD83D\uDDE3 Discussion topics
Time
Item | Presenter | Notes |
---|---|---|
Codejail | Is 2U planning to use the edunext service or build out a different one that’s django based?
| |
Retro of DEPR Breaking Changes Pilot |
| |
Teak Maintenance Priorities |
|
✅ Action items
- Kyle McCormick Will open a discussion post about DEPR, breaking changes, etc. Include link to pilot updates in open-edx-proposals (6 month window)
- Feanil Patel will bring up pilot with the DEPR WG and see if we can land an OEP update.
⏺️ Recording and Transcript
Recording: https://drive.google.com/file/d/143ymL_DZFW46QZVTOR2_skrHGRAXW5Il/view?usp=sharing
Expand | ||
---|---|---|
| ||
Maintenance Working Group Meeting - 2025/01/09 08:58 EST - TranscriptAttendeesAdolfo Brandes, Feanil Patel, Jeremy Ristau, Kyle McCormick, Michelle Philbrick, Robert Raposa, Sarina Canelake TranscriptFeanil Patel: How's everybody doing? Robert Raposa: Okay. Kyle McCormick: Good morning. Quite well. Kyle McCormick: avoid the constant merge conflicts that we get in the text files. This is an idol using want to work on. Feanil Patel: Yeah, I'm not sure,… Feanil Patel: but it is on my list to check out the landscape of Python environment. government management has changed a lot in the last year and… Kyle McCormick: Yeah. the requirements text files. Feanil Patel: It is on my list to sort of experiment with and try to see if we want to do something different. Robert Raposa: Where are the conflicts happening? Sorry, just curious. Feanil Patel: Yeah, because we do the weekly requirements merge and so if you have a PR… Robert Raposa: And does that include even with the action of that that targets a single requirement? Yeah. Feanil Patel: where you're changing requirements then you're constantly having to rebase it. Yeah. Kyle McCormick: It can depends on the thing you're upgrading. some version bumps will hit a lot of secondary dependencies. Kyle McCormick: Although possibly I think I've been struggling to run the upgrade one package locally. If I can get that working, then maybe that'll solve lots of the problem for me… Feanil Patel: Yeah. You can Yeah,… Kyle McCormick: because I don't always love waiting go through going through the GitHub workflow says Robert Raposa: I think I don't know… Robert Raposa: if it's an internal doc so I can find but we have just manual instructions for easily being able to do that. and it might not be running the GitHub action. It might just be like adjusting make file a tiny bit and then just running exactly. Kyle McCormick: That works. I'll try that next time. Feanil Patel: I do that all the time. So All right,… Kyle McCormick: Hey, Jeremy. Feanil Patel: to our actual meeting. Welcome to 2025. It's the future. Still no flying cars, but here we are. I think we have a bunch of dudes from last year that we should go through and then we have two topics, but I suspect one will be a decent discussion, the other one hopefully will be pretty quick. so let's kick us off with Kyle. There's an ADR around the settings file stuff. Is that still on your plate? You're muted. Kyle McCormick: It is still on my plate. as we were discussing in our team Slack, I'm going to try to make a crack at that next week and… Feanil Patel: Yeah. Yeah. Kyle McCormick: have a ride up available for everybody. Definitely. Feanil Patel: I'm excited to help you with that. I think making those decisions and making the design clear will help and maybe we can play around with trying out some of the design choices and seeing how it would look. and then Jeremy, do you have any updates on this performance testing on the deer for the forums? Is that still relevant to you? Jeremy Ristau: in so much as the requests to do performance testing and… Jeremy Ristau: trying to get the specifics of what our traffic sort of needs. that request has been given to the Infinity team and I know that they're doing a lot of testing right now. and they're logging issues and updating things. So, … Feanil Patel: Okay. Feanil Patel: Sounds like this is in progress and you don't need it tracked here anymore. Yeah. Okay. Jeremy Ristau: that's probably the right way of saying that. Yeah. But just, caveing that, checking it off here doesn't necessarily mean that it's all Feanil Patel: Yeah. Yeah. It's It's not done, but it's like being and then my followup with Ed and Felipe sort of devolved into but what if we jumped even further into using W wasam to run code jail. So I'm sort of trying to pull that back into a conversation that's great but what if we just landed the thing we have now instead of this aspirational future. so which I think we will arrive at, but I just had to undetour a little bit from the last time we had that conversation. 00:05:00Feanil Patel: But Robert, do you know if are you guys using the EDX service for your containerization stuff or what are you doing for that Yeah. Robert Raposa: So, we're going to figure that out shortly. Robert Raposa: And I don't know if we're going to actually get a Django version of it going or I don't know what we're going to do yet. And we'll probably know about that shortly. So I will let you know what our intent is and whatever it is could be an openex service or it could be the one that exists. basically I know Tim had thoughts about the service that does exist and… Feanil Patel: Yeah. Yeah. Robert Raposa: I don't know what his thoughts are. Feanil Patel: Yeah. I would love to get those thoughts. Robert Raposa: So yeah,… Feanil Patel: Because maybe I think this is a space for collaboration. Robert Raposa: exactly I will definitely share shortly. Feanil Patel: Okay, and then I'm going to try to do probably the week after realistically. and then deer tickets for frontends that were deleted. Is that still on the deep backlog, Jeremy? To get right. Jeremy Ristau: I mean the problem is they're not really delete they're not ready for deprecation on our side. So it's the work to roll them out and then the writing of the deprecation is right after that. Feanil Patel: Yeah. Yeah. Yeah. Okay. Jeremy Ristau: So I wouldn't say deep in the backlog… Feanil Patel: Yes. this is a situation… Jeremy Ristau: but I would say comfortably there. edge. Feanil Patel: where from your perspective you need to prioritize internally getting them onto edge before you can turn them off. Jeremy Ristau: Yeah. Yeah. Which is just we don't have a lot of people to offer feedback and… Feanil Patel: Yeah yeah yeah got it. Jeremy Ristau: it's just a hard word to dep prior to prioritize here. Feanil Patel: |
...
Feanil Patel: Yeah. It is something that I think I know Kyle and I have been talking about this more and we'll probably talk about it more today in the deer meeting, but I think we're getting close to starting that work ourselves. how is that going to impact and what do you guys need from that? Because we can't deploy it to edge for you certainly. Jeremy Ristau: Yeah. … Jeremy Ristau: while I don't recommend this as just like standard operating procedure, external target dates are a means of prioritization. While I don't recommend that standard operating procedure. Feanil Patel: That's the best news I ever heard. Feanil Patel: Yeah. Yeah. I don't want to that is not my favorite hammer either. Jeremy Ristau: Yeah. Yeah. Feanil Patel: And also I don't want to use it multiple we already talked about it earlier this week so I don't want to use it multiple times all at once… Jeremy Ristau: Yeah. Exactly. Yeah. Jeremy Ristau: Yeah. Awesome. Feanil Patel: but okay that's good to know. Jeremy Ristau: Good stuff. Feanil Patel: Just want to understand if that's going to block further or if we can sort of move it forward. cool. Jeremy Ristau: Yeah. I don't see this as complex, … Feanil Patel: Yeah yes yeah. Jeremy Ristau: large scale thing. It's just like getting a team to do that is tough from a product perspective. Feanil Patel: Yeah. Got it. Jeremy Ristau: So, yeah. Feanil Patel: Understand that. All and then I think the biggest conversation for today is actually to do a little retro on the deer process pilot that we've sort of been playing around with over the last year. I'll post this in the messages here so people can jump straight to it. and I can take some of the notes. Just pull it up myself. to recap, we've tried a couple of different variations of advanced notice for breaking changes via the deer process. Feanil Patel: In the end, I think what we've arrived at right now is notifying about when things are going to be breaking is sort of the critical piece,… 00:10:00Feanil Patel: not guaranteeing that we don't make the announcement. I don't know. How do I say this in a succinct Kyle, do you want to give it a shot? Kyle McCormick: Yeah. … Kyle McCormick: the current system is every breaking change goes through the deer process and that you have to give a month heads up or hey, so you're supposed to give a six-month heads up by default that it's going to happen. and there needs to be a one-mon window of overlapping support when that is possible. So if you're replacing 6 months in advance X is going to get replaced with Five months later at maximum you introduce Y. So you have functioning system with both X and Y. And then a month after that you can remove Does that jive with everyone's understanding of the deer pilot as we have it today? Kyle McCormick: Right. That is the Robert Raposa: Yes, with the one addition that there's no one rule fits all scenario. So it's also be like and we're going to reasonably discuss any individual thing like hey… Feanil Patel: Right. Yeah. Robert Raposa: if this change is actually going to take three months to implement we would give more time for it or whatever. So that can't be it. But otherwise, it was a great summary. Feanil Patel: So, I think it's that one month of overlapping window of support. and I think these the ideals, is ideally six months of warning and one month of overlapping support. Often we find shoot, this thing is got to get upgraded very soon. We're going to do it right now. and we will leave overlapping support for… Kyle McCormick: Before we jump into solution space,… Feanil Patel: what seems reasonable. Kyle McCormick: would it be worth and just starting in a what's going what's not going? Feanil Patel: Yeah, let's do it. Kyle McCormick: I want to make sure everyone gets the chance to voice their opinion around this before we start discussing stuff. Feanil Patel: So, yeah. What's working and what's not working? one thing that I think is not working is doing it for every single change is fairly costly and it's hard to Robert Raposa: What is the it and… Robert Raposa: what is the cost? Do you mean the warning do you mean creating a deber ticket and announcing it or Feanil Patel: Yeah. Yeah. Feanil Patel: And I think this is a thing I think it's fine honestly, but I do want to sort of acknowledge that this if we have to make this for every single breaking change, we're going to miss some. but we do our best and I think that's what the deer process is for. Jeremy Ristau: just because I feel like that was an open question. Are you able to clarify which parts of what Kyle described are the points of most friction or points of most cost? And then what is that cost? Is it statement of work dollars? Is it, things you can fit into a week? can you be a little bit more explicit? Feanil Patel: I can be a little more specific. Yeah. I think in particular it is sort of the mental overhead of knowing when to use the process and the number of times that you have to actually do the write up and go through the process that's costly. And so that's like if there is a change happening the mental overhead of does this change need to go through the deer? Is it actually something that everybody needs to know about or is it like a refactor in an internal system and both the thing that depends on it and the thing that we are changing are under our control. So is this an interface or not is a question that's very difficult and requires sort of a lot of upfront thinking. 00:15:00Sarina Canelake: Yeah, something that I'll add to just in terms of cost is that I think the people who attend this meeting regularly are pretty aligned as to what the process has been and has been willing to iterate. But I'm thinking of a couple of times Peter Pinch has been like, I don't understand this process at all. I don't understand what I need to do. I need to be totally walked through it. And I think Peter's a reasonably engaged community member. so I just wonder what the cost is for people who might want to engage in this type of work but it's deprecation work and who are not necessarily a part of this brain trust at this meeting? Feanil Patel: Did it. Sarina Canelake: Again I don't know if I place a value judgment on that. I just think that that's something to think about is how accessible is a process to a average community member and… Kyle McCormick: Yeah. … Feanil Patel: Yeah. Heat. Sarina Canelake: how much do we help along with that? Kyle McCormick: Phineal touched on a point which is what goes through this process. If we're defining this process as for every breaking change, it begs the question of what is a breaking change? If you change a method Pi somewhere in EdX platform like change the method signature you add an argument by some definitions I think dayworms we would consider that a breaking change like you said to that effect if every time we add an argument to a function in an API.py Kyle McCormick: PI file and IX platform we need a deer ticket and the deer ticket is going to be phrased something like we're removing support for the version of this method that doesn't have this argument that's crazy and that's my problem with I guess my more sinly stated is to follow the process to the letter of the law requires a level of conscientiousness that I don't think anyone here is capable of or willing to do so we have been following at a lower level of conscientiousness, which is fine, but I think we should define it Feanil Patel: Yeah. Yeah. Yeah. Robert Raposa: Yeah, I imagine this is one of the places where it's going to be impossible to write anything that is black and white and solves everyone's problem, so we can, provide examples and give our best guess as to what is a good way and not a good way to handle certain things. and there is a separate question for API changes of the nature that you're talking about. If it's not the deer process, is there any way to actually list those as breaking changes in a place where someone could actually read about them? Is it just follow along with commits that have exclamation marks, and if that's what you want to know and that's how you can find that information. Robert Raposa: And this is just for things that are,… Feanil Patel: Yeah. I think the other part of Kyle's point that I want to sort of emphasize is this the way that we have to define something as a deer is different from the way that a breaking change is communicated. Robert Raposa: a little bit more potent. but it did come up with Peter Pinch's example of hey, I want to remove something that is broken. Do I need to follow the deer process? and we ended up saying yes. And maybe that was not the right answer. Who knows? and Feanil Patel: the breaking change this changed from A to B. Here's your transition. Whereas a deperator is often this is going away. and so we end up writing sort of these convoluted phrases like removing support for using forks of the footer as a way of configuring the footer, which is a really confusing sentence. All of which is to say forks of the footer will no longer be guaranteed to work because we're changing a bunch of stuff with how the front end is put together, right? Feanil Patel: And so I guess one question is it that we're trying to put shove breaking changes which are notifications into deer which is the process of not just communication but of acceptance and review and should there be sort of a separation of we're going to be changing things all the time. Those are communicated breaking changes. and then when we're getting rid of something that has been relied upon externally by a lot of people, we have a deer. Or is it that we can have deppers and deppers can be just sort of like all kinds of different things? 00:20:00Kyle McCormick: I'm gonna dodge your question for so we can keep talking about problems. But yeah, Feanil Patel: Sound good? Yeah. Michelle Philbrick: I'm a non-technical person, but just listening to this, I don't know if this would be extra work or headache, but could you categorize them? one is a deer that affects everybody. it's going to take a while on the reverse side maybe one is the lowest we can just communicate out that this is happening but again I guess it goes back to who decides I don't know if that would be this group but I don't know if that would help. Michelle Philbrick: So people at least can see we are aware that this is happening and here's where it is in the priorities. And if none of that makes sense,… Feanil Patel: Yeah, that makes sense. Michelle Philbrick: just disregard this. Feanil Patel: No, I think that makes sense. I also think that this might be the sort of thing where us investing in a rubric for when things need lots of processing. Kyle McCormick: solution Stop it. Feanil Patel: All Fine. I'll stop. I can't help it. Kyle McCormick: I know. Feanil Patel: I will say a thing that is working real quick before you hop in there is I think a thing that is working is that people are going to deer to find out what might break them and that process is understood by a large swath of the community. and they know that that's where they need to look and I think that is really useful. Sorry, go ahead Jeremy. Jeremy Ristau: expansion and contraction model of maintenance and using that as sort of the way of thinking about things. I think that's good from an operator's perspective. Maybe not from a speed perspective of master but from an operator's perspective. I think that's a good mindset to have around it. I think that leveraging a communication medium that people pay attention to was smart at first. something that's core and not going to just be ignored so we can do whatever we want with it and overengineer and nobody feels the pain because nobody looks at it. Jeremy Ristau: one thing that I think with some of the recent stuff I would say that maybe didn't work was the six months I think is supposed to be like the hey go get ready, you've got six months and at least for us there was confusion around when we could start getting ready cuz there's a lot of work that's happening maybe not necessarily something that we can absorb immediately and start to get ready for and then that is not 6 months then maybe it's only 3 months or it's only two months and so I think the spirit of what we were trying to go for with six months in practicality ended up maybe not being at that much of a buffering window Jeremy Ristau: that's what I got in my head right now. Feanil Patel: I feel like that was an overloaded idea from because it was one of the intentions was hey you should tell your product people that you're going to need engineering hours six months from now and that's plenty of warning for them but another was like hey you should figure out how this change is going to impact you and do your internal research about how your systems will be impacted by this change between now and six months from now and that one is I much harder because when we announce we're going to get rid of a thing, we don't necessarily have enough detail for that internal sort of operational work to begin. 00:25:00Kyle McCormick: Yeah, the removal process uncovers some of the breaking changes. Feanil Patel: Right. Right. Jeremy Ristau: Yeah, Yeah. Right. Feanil Patel: And so it's like only that last overlapping month where you really have a high confidence thing that you can do technical research against and then you have to sort of scope the work and do it pretty fast. Jeremy Ristau: Exactly. And we're choosing month, but it could be two months, it could be three months,… Feanil Patel: Yeah. Yeah. Jeremy Ristau: but that window might happen with five months left or it might happen with two months left. And the moving target is I think stressful. Yeah. Feanil Patel: And on the flip side, I think it's like I'm sorry. Go ahead, Robert. Yeah. Robert Raposa: No, go ahead because I was going to just make a different note to respond. Feanil Patel: I was going to say I think that it's one of these tricky things where it's like as an engineer, I'm like, " I know this file needs to be deleted. I just want to delete it as soon as possible." but from a process perspective, it's I think the best time to do this is on February 12th and not a day before. Right. Red. Yeah. Jeremy Ristau: post removal of said file and yeah I know the discovery and experimentation portion of these kinds of works can be very tricky yeah… Feanil Patel: And if you start adding calendar time to it that is not just the engineering time then it becomes really long and you start losing context and the costs go up. So that's a hard thing. Jeremy Ristau: then the longer it goes the more people you have to bring in and then the longer it takes then it just it snowballs for sure Yeah. Feanil Patel: So I think that's a balance that we have to sort of constantly be struggling with. Jeremy Ristau: And I think that gets back to I know we're not solutioning yet,… Jeremy Ristau: but we can't put everything through this process because of the bloat that comes along with it. And so where do we draw the line about when to introduce the overhead because the overhead is worth it? Feanil Patel: right. Yeah. Kyle McCormick: Yeah, and to restate that particularly for certain operators that I've heard from they're watching Deer because they want to hear that notifier is going away. That's a whole Django service is going away. They're looking for that level of chains and when their feed is inundated with I'm trying to look for a good one but API function changed or we're removing node 18 support from this particular front end app that makes the whole process less useful for them if there's no sort of filtering in place that they can use. Feanil Patel: David, did you have a different thing you wanted to say? Kyle McCormick: I don't know if you're talking about blow in terms of the input or the people consuming, but I think it goes in both directions. Robert Raposa: believe it is it's just to decide whether or not it needs to be deer or not. As if deer is a super definfined thing. The whole point of what we're doing is trying to figure out what the deer process is. So, to Michelle's point or whatever, it could be like, here's a simple deer and here's a more complex deer but there's still lots of things about the deer process, which is simply an expansion and a contraction and notification that is useful, Kyle McCormick: I think we're retroing the whole right now. Robert Raposa: Right. Kyle McCormick: Deer and Robert Raposa: But I mean I was going to say that before we did this pilot, I had a lot of concern about it. Feanil Patel: terrifying. Robert Raposa: It was unclear how many deppers were being created and that they were all going to land on some target, release and then there was just going to be a swarm of deletions that could all happen on the same day that just made no that was very potentially anxietyinducing. And I feel like this pilot has totally changed that for me, So there may be lots of iterations of improving and tweaking for handling different things, but it's so different than that world. And so I'm much more appreciative of the pilot than what it was like before we did any of this. So wanted to make that 00:30:00Feanil Patel: It's 9:30 technically I think that is the end of the general maintenance meeting and… Feanil Patel: there's an edex platform portion are there edexplatform topics would people like to continue this conversation and move into some solution space to people want to talk more about concerns here. Kyle McCormick: Can I propose that rather than trying to come up with a fixed process by the end of this meeting,… Kyle McCormick: we take these learnings and we go back and write proposals for a future maintenance Okay, cool. Feanil Patel: Yeah, I think that's sounds fine to me. are there other thoughts? I do think that one of the things that really did work was having not just a release target but a month target for when we want to land things. I think that sort of specific tactical change helped a lot. Kyle McCormick: I agree. I think a wish of mine is that the deer process, whatever the breaking change process is, if that's deer or not, and the product process, if we had some really I want a contributor who's not in this meeting to understand which of those processes they need to use and how to use them for any given change and them to be kind of yolked together in some comprehensible Jeremy Ristau: I have to chime in plus one when I heard Phineal originally talking about sort of we don't know when to do it and other feedback I've heard this same feedback in to you about whether or not a change should be going through the product proposal process. and I think it's that the clarity around when to send it down this path or when to send it down this path or when you don't need a path. and you can just right really I think that's a big point of confusion right now. Kyle McCormick: Yeah, I'm also going to throw hops and ADRs into that as processes that we have depending on the situation. Jeremy Ristau: Yeah. Yeah. Feanil Patel: or what? Kyle McCormick: … Feanil Patel: Which process did you say? Kyle McCormick: OAPS and ADRs Feanil Patel: Yeah. Yeah. Jeremy Ristau: There's so much subjective sort of in the- moment choice making. Kyle McCormick: Step one, work on openex for eight years. Step two, you'll know which one to use. Jeremy Ristau: But you might still get it wrong in that. Feanil Patel: Yeah. Yeah. Kyle McCormick: Literally every time I get it wrong,… Feanil Patel: It's like you might still get it wrong, but it's okay because you've worked on it for eight years. So, people are very forgiving because they know Yikes. Yeah, we'll make it better. Kyle McCormick: new contributors watching this video, we see you. We feel for you. Feanil Patel: Yeah, we're trying to make it all right. Let's do proposals and maybe we can talk through those proposals next time. the other thing I had was I think there are in fact not any teak maintenance priorities and I just wanted to make sure that everybody else also believes this. I think that what is you Elmo? we have some Elmo priorities because we got to land Django for Almo. Feanil Patel: But nothing that is specific to teik. we can start on the Django work now. but it's not critical to te and I think that's okay. Adelf I'm kind of looking at you to figure out… Feanil Patel: if there's any front end related if there's yet another node upgrade we need to do for te or react or whatever. that's a good one Jeremy. Yeah. Yeah. Jeremy Ristau: Yeah. Maybe not like a whole thing,… Jeremy Ristau: but that last piece of that thing. Feanil Patel: Yeah. Yeah. Jeremy Ristau: Yeah. Yeah. Feanil Patel: That's true. Yeah. That's probably worth mentioning,… Feanil Patel: too, because operator wise, it's pretty high impact. Delta refuses to answer my question. Sarina Canelake: So, I saw something fly by yesterday. Sarina Canelake: I'm trying to find what chat room it was in. that Braden had said something about maybe getting his React 18 upgrade for Paragon over the line. Sarina Canelake: So I don't know… Feanil Patel: Yeah, I did see that. Sarina Canelake: if that's something that's at least worth tracking. Feanil Patel: And I think whether there's downstream impact of that was something I was hoping Adela would tell us instead of rebooting his computer. Sarina Canelake: Yeah. It's not that he refused. Looked like his microphone was maybe not working. 00:35:00Feanil Patel: Yeah. No, it looked like his computer just sort of fell off the face of the earth and he'll be back as soon as he can. Sarina Canelake: Yeah,… Feanil Patel: I don't think our Delta is avoiding it. Sarina Canelake: we could follow up with it. It sounds like actually Brian Smith was really helping Braden on this,… Feanil Patel: Yeah. Yeah. Yeah. Hey,… Sarina Canelake: so maybe it's worth following up with Brian. Feanil Patel: I can hear you. Adolfo Brandes: Yeah, sorry. Adolfo Brandes: Issues, technical issues. Feanil Patel: Very good. Adolfo Brandes: There's less maintenance stuff and more product stuff on the front end side happening for TIG. Adolfo Brandes: So, I'm not aware of any big other than new thing that we have to do or So we have a plan that entails I don't know… Feanil Patel: Yeah. Yeah. Feanil Patel: As there was the Braden's React upgrade in Paragon? Is that going to require Paragon? I guess it doesn't require people to upgrade Paragon. It just allows people to upgrade to a nicer version of Paragon. Adolfo Brandes: if this is maintenance or a new thing but where we want to standardize dependencies across the front end right I guess it's half maintenance half Feanil Patel: Right. Absolutely. Adolfo Brandes: new in the sense that we want to keep them synchronized going forward. and there's a hope that we will be able to do this by te and it includes possibly upgrading the MFS to use front-end base as a unified library… Feanil Patel: Right. Yeah. Adolfo Brandes: but not necessarily as via the shell, so it's just a mechanism to keep all the dependencies either in a single place or synchronized right so that's the teak thing I have in mind right now but I don't know if we're going to be successful by teiker refactor Feanil Patel: And that's sort of refactoring improvement, not necessarily a breaking change. The idea behind that change is that operators don't need to deal with something as a result of that change. So that Yeah. Adolfo Brandes: Yeah. Yep. Feanil Patel: Yeah. In which case I think it doesn't sort of bubble up to what this group needs to sort of help manage or maintain. I think we do support somebody put I thought we already added 311 edex sandbox. I think we do support 311 but I think we officially dropped support for 38 which I think we haven't communicated out and we just need to communicate out that 38 support is gone. Kyle McCormick: I Okay. Feanil Patel: I thought we did that already but I need to go look and… Kyle McCormick: Yeah. Yeah. Feanil Patel: Feanil Patel: Yeah. Kyle McCormick: That read me there says we've already done it whether this gets back to communication. Feanil Patel: Okay. Yeah. Yeah. Yeah. Kyle McCormick: Could there have been a deer ticket for code doesn't support Python 3.8 even… Feanil Patel: Yeah. We have the whole Python 3 38 up. Kyle McCormick: though we had the we didn't because we weren't doing the pilot at that point. Feanil Patel: That's true. We had a Python 311 upgrade ticket, but we didn't have a 38 deer. Really? Cool. Kyle McCormick: So, this is pertinent. When you're writing your proposals for what to do about Deborah and breaking changes this week, everyone think how would you like to communicate hear about dropping of language version and framework version support? Feanil Patel: That works. Yeah, we'll come back to that. But okay, it sounds like we should figure out what the situation is with the code stuff and make sure that everybody is on the same page, but other than that there aren't really any major things for teik, which is great. and we'll start focusing on the Django upgrade for Almo because I think we can in the libraries at least start adding support for running both 42 and if 52 doesn't come out for another month or two is the real problem. Feanil Patel: we can start testing. I think that's what we're waiting for with the Django stuff is for 52 to be available and once it's available, people can start upgrading everything all at once rather than spending some time upgrading to 50 now and then 52 later. Reduce operator and maintainer burden a bit, even if it shrinks our sort of time line a little. Okay. Anybody have anything else? Kyle McCormick: How long is one week good for making proposals about deer or do folks need longer? Feanil Patel: I also want to remind you, Kle, that next week we're in person, so you and I might not be able to make this meeting as easily. 00:40:00Kyle McCormick: Can we do two weeks? Feanil Patel: Feanil Patel: Yeah. Yeah. yeah,… Kyle McCormick: Cool. Thanks, Feanil Patel: maybe we should just cancel these meetings for next week unless anybody here has objections because four out of the seven of us are not going to be out of the seven of us. Yeah. Robert Raposa: Sorry, I was not going to talk about the schedule on time with any schedule, but in terms of proposals, I guess I'm a little confused about the process that we're looking for because it wasn't really for me it was more like, hey, we were piloting something. Kyle McCormick: Yeah. Robert Raposa: Is the pilot better than what we had? And simply and it's what we've been doing? Do we simply want to update the documentation to say here's what we're doing and we can endlessly improve upon that and if people want to put together proposals and improve upon that let's do that let's not do that but to pretend I just wanted to get out of the place where we've got a deer OAP that states the process and… Robert Raposa: we've got a pilot that we're pretending is not the process even though it is the process and can we just bring those things together and I'm unclear if we're saying yes we can or No, we can't. Or now we're going to write new proposals of improved things and have a new pilot. That's also not the comment,… Feanil Patel: Yeah, I think this comment by you,… Feanil Patel: Robert, kind of very succinctly summarizes the place we are currently at and should go into the O like as soon as possible. Robert Raposa: but and so we're okay. We're accepting the pilot at that level, right? And… Feanil Patel: I think so. Robert Raposa: then yes. Feanil Patel: And then we can make more changes. What do you think, Kyle? Feanil Patel: Yeah. Yeah. Kyle McCormick: has been working for us and I'm cool with encoding that. I think in terms of proposals,… Feanil Patel: Yeah. I think we can say Yeah. Kyle McCormick: I think what do we do about breaking changes is still worth addressing. yeah, sorry. Feanil Patel: Feanil Patel: I think that makes sense. Yeah. I think the timing is really useful and it would be good to put it into the OAP so that other people can more easily find it and understand what they should do. when they should do it I think is still a hard question and probably one that we can improve. There are a lot of ways to improve how people determine what needs to go through the deer process. But we can still say if you're going through the deer process and it is a breaking change. here's how you should do it minus negotiation. Feanil Patel: And we can continue the conversation in this group about what level of change or maybe even there's maybe a better conversation to have in the deer working group about what level of change what should merit a deer and how can we make that clearer and… Feanil Patel: easier for everybody. Kyle McCormick: I Yes. Kyle McCormick: And also I want it to be a coherent story with breaking changes in general and with the product process… Feanil Patel: Yes. Yeah. Kyle McCormick: which having Jeremy for is really really useful and I know Jeremy doesn't go to Denver. So it's part of why I've been having these conversations here. Feanil Patel: Yeah. Yeah. Feanil Patel: Yeah. That sounds fine. in two weeks let's loop back and talk through propos maybe focus more on how do we determine what can we do to make it easier to know when to use the deer process because I think that's sort of the fourth thing we're trying to figure out. Kyle McCormick: I could not answer that without answering the breaking change question and… Kyle McCormick: the product process question. Feanil Patel: Yeah. Yeah,… Feanil Patel: I think that's right. Yeah, the timing. Robert Raposa: gives more focus to what you're looking like when you were saying everyone put together your proposals over the next two weeks. it's really that question and not about the pilot and all the dates and other things. I think yeah and… Feanil Patel: Yeah. Yeah. Yeah. There's a lot of aspects that work that we should encode. And then it's still hard to know when to use deer. I think yeah,… Robert Raposa: we can even throw up in a discussion topic that's like hey we're having difficulty with this people throw out ideas here or write your proposal or whatever this is what we want to improve. Kyle McCormick: Yeah, I can take on that discussion topic. Feanil Patel: let's do it as a post. Let's see if we can get the community involved and help It'll also sort of bubble the topic up in their mind which I think will be useful to get them to use the process more sense. 00:45:00Kyle McCormick: Yeah. Yeah. Jeremy Ristau: This is a public post. Is it also going to include these are the parts we're encoding because we thought they worked out really well. Yeah. Okay. Jeremy Ristau: Pilot updates. Feanil Patel: Yeah. Yeah. Jeremy Ristau: I like that. Robert Raposa: I mean that Yeah. Feanil Patel: To do it. Kyle McCormick: And if some I can do that once someone makes the PR to open edex proposals. That way I can just say here's what we are doing. Here's the PR. Kyle McCormick: Feel free to go comment on it. here's what we still need to solve. Feanil Patel: Robert, would you mind making that PR? Robert Raposa: I may be able to do it one I was going to make a related process question… Robert Raposa: which is it may be a very short thing that can be done as part of the deer working group that I can't be at today. and it could be like, hey, we're literally just pair program, write the little paragraph in the right place and just approve it and just get it done all in as a possibility and it should be discussed there anyway. Robert Raposa: And… Feanil Patel: Yeah. Yeah. Robert Raposa: then it'll be clarifying here's what we're actually changing and I mean the deer working group should know about the deer o updates. and you'll let me know I can't Yeah. Feanil Patel: Yeah. Why don't I take that? Yeah. Yeah. Yeah. Robert Raposa: Robert Raposa: So I can't be there. You'll let me know how my proposal that I just gave goes after the deer meeting and then if not discuss… Robert Raposa: what can happen. Feanil Patel: Yeah. … Feanil Patel: hopefully any concerns we will push onto that deer ticket so that you can see it all there. But ideally we can just take I think this two paragraphs here and try to put them into the deer opp today at the working group and… Feanil Patel: we can figure it Cool. All right, I think that's everything. Robert Raposa: It is 9:47. Robert Raposa: We have a little extra time. So, we could do what I'm proposing happen there right now. Feanil Patel: Just like you. Robert Raposa: Whoever wants to stay on and… Feanil Patel: Yeah, we want to do it. Robert Raposa: we just are like, "Okay, let's taste it and… Feanil Patel: Yeah, let's do it. Robert Raposa: get it done and agree on it." because that's really what I was mentioning happening. Feanil Patel: Do you want to share your screen? I'm going to stop the recording. Thank you everybody for coming. Feanil Patel: look for the PR. Robert Raposa: Thank you. Meeting ended after 00:47:49 👋This editable transcript was computer generated and might contain errors. People can also change the text after it was created. |