2023 06 28 Sawtooth Community Call
Ry Jones: You have 4. Kevin O'Donnell: Okay? So anyone's not familiar with it already. There's the antitrust policy and the code of conduct policy. If you review it. and make sure you follow it. Kevin O'Donnell: So just following the Kevin O'Donnell: the basic agenda that I put out last time and gonna run through everything Kevin O'Donnell: So first step is the view of what we said we do last meeting. Kevin O'Donnell: But it was a Tsc update. Thank you, ry, for finally actually submitting that Kevin O'Donnell: for James. They the outstanding questions. I took a quick look through where it is. Both of them were like grammar and spelling things, I think. I think all the sort of substance of issues are going to report. But I will take another look at that Kevin O'Donnell: night. What is the in terms of the process. Is it like it has to be a unanimous improve?
Ry Jones: No, usually. it's 2 weeks after submitted. If there are any outstanding questions. Ry Jones: One of the larger outstanding questions was, the Maintainer's list is way out of whack Kevin O'Donnell: right on me. Ry Jones: So that's Ry Jones: Peter is also on the to, so I'll let him speak to that. But, Peter, was there anything else that you recall from the last Doc meeting about that? Peter Somogyvari: Yeah, I figured the Peter Somogyvari: someone. I don't remember who it was, but someone brought it up. That
Peter Somogyvari: it seems not very up to date that there were people, Lisa is maintained. We haven't done Peter Somogyvari: anything on the project for maybe 6 months. So I think what they were saying is that maybe that could use an update to double check who is still actually active as a Maintainer. Kevin O'Donnell: Right? The so agreed. The certainly can be printed out. I think the question a little bit is Kevin O'Donnell: whether anybody needs to come into as replacements for stuff.
Kevin O'Donnell: the Kevin O'Donnell: principal sort of active maintainer really are are Andy yourself, and from right. I don't see very many of the others from the existing list go through and do active reuse emerging. Andi Gunderson: Yeah, that's pretty accurate. Andi Gunderson: we have in the past. Andi Gunderson: when someone has asked to be removed, gone through and sent out emails to people to ask them if they want it to be removed. So there's some people on there who have actively said they wanted to stay on and sometimes pop up the example of that, like Shawn Montgomery? who was very active in the project originally. But we haven't done that for a while. Kevin O'Donnell: Okay. what should be the process? We deal with that in terms of, should we go and do that again? Or should we just take it? By the the which I mean, there is a rule right with of Kevin O'Donnell: so many months. It active you're supposed to be
Kevin O'Donnell: was it? Retired and then. Kevin O'Donnell: and pushed off? Or should we go through that process again see if anybody wants to. Then Kevin O'Donnell: actually, it's probably better to actually set up the emails as we just them and try to get them more involved again or explicitly out we get. I, my thinking is, give one more thought.
Andi Gunderson: Yeah, so sorry. Go ahead. I just wanted to say, Andi Gunderson: Shawn is unfortunately on vacation right now. Why not on the call? But we had talked about going. We have Andi Gunderson: The Maintainer rules in the Andi Gunderson: so I I forget what reports in the Andi Gunderson: r of T repo. Andi Gunderson: yeah, that's it. Sorry not fully awake yet. Andi Gunderson: And we had talked about going through and revising that to not only make it Andi Gunderson: easier to know when we should retire people, but also make it Andi Gunderson: clearer way. We would move someone into Maintainer Status. Andi Gunderson: both of those have been. It's it's pretty like up to the Current Maintainer group. So I obviously, as the Maintainer Group has gotten smaller. That has become a little bit more
Andi Gunderson: work. since we haven't had time to really deal with it recently. Andi Gunderson: so that's one thing is make going into that and updating the rules. So they're really explicit. it'll also help us get new more maintainers from the perspective of it's very clear what someone would need to do before we could make them a maintainer. Andi Gunderson: unfortunately, we don't have that yet. Kevin O'Donnell: So what would be the next step? Just we gotta wait for Shawn to come back. Kevin O'Donnell: They have that discussion and put it forward. Or is there something we can do now? Kevin O'Donnell: I guess that in terms of moving on moving things along Kevin O'Donnell: when it. When is Sean do back
Andi Gunderson: next week? Kevin O'Donnell: Next week Ry Jones: I do collect stats on maintainer activity. Ry Jones: or actually, I collect stats on all activity. Ry Jones: So it's pretty easy for me to get that data. Kevin O'Donnell: hmm! Ry Jones: I can do it again. Kevin O'Donnell: So I'm putting down.
Kevin O'Donnell: Make sure to update notes for this. So we need a a refresh of the container rules rules. We need to Kevin O'Donnell: to a poll of the existing maintenance. Kevin O'Donnell: Yes. Peter Somogyvari: one more thing that I can recommend is Peter Somogyvari: if someone wants to stay on, but they haven't done anything in a while. Then you could make them maintain your emeralds and just list them in the same document, but on a separate list. So that way Peter Somogyvari: they do get the credit for life that they've been putting work into the project which they should. Peter Somogyvari: but at the same time it's not confusing. It doesn't look like there's a bunch of Maintainer when there is.
Ry Jones: Yeah. I actually had a bunch of pull requests to do exactly that a while ago, and they were not received. Well, so Ry Jones: I would like it if, instead of removing Maintainer in active ones, we're moved to emeritus. And then it makes life a lot easier. Kevin O'Donnell: I mean, I would prefer that, too. Right? When I see a Maintainer's list, I expect somebody actually to be. Kevin O'Donnell: they're taking a project right? Peter Somogyvari: Yeah. And then, if you keep their names on there. But in the Peter Somogyvari: I think that's a very reasonable trade off that Peter Somogyvari: they should be. Everyone should be happy to make. Peter Somogyvari: There's their navy still there, but they're not any more actively maintaining.
Peter Somogyvari: So if it was me. Peter Somogyvari: I wouldn't bother too much with the poll either. To be honest, I would just look at the stats, and Kevin O'Donnell: I would say, I think there's a reason to go and and and at least contact everybody to see if they're Kevin O'Donnell: willing to come back into the and Kevin O'Donnell: get active on it. Maybe even just give an update because it's it's entirely possible that they just haven't been paying attention to what's happening here. Kevin O'Donnell: but I think I think it's still worth the effort to go around and see if anyone still interested in call.
Peter Somogyvari: If you have the time. Kevin O'Donnell: Time is there to be spent? Kevin O'Donnell: Okay. the okay. So that's the the next update is due Kevin O'Donnell: the lie. Kevin O'Donnell: Try to do better this time. Kevin O'Donnell: is it due July, or is it due to this? Is it already due? Kevin O'Donnell: I would have to look at the Kevin O'Donnell: The last one was for April so quarterly so soon. July. Ry Jones: okay. So
Kevin O'Donnell: I want a volunteer from my guys to help me with the Poc report Kevin O'Donnell: or anybody else. Kevin O'Donnell: I have a feeling. It's going to follow me to end up doing it this time. The mark you want to help me out with that. Mark Carroll: Oh. Kevin O'Donnell: it's basically just a pull request of a mark down document against the Kevin O'Donnell: against the Tsc, just to say, what's been happening in the product.
Kevin O'Donnell: Okay. Kevin O'Donnell: it's it's very formulated. Really. Mark Carroll: Okay.
Ry Jones: So saw tooth is July 20 seventh. Kevin O'Donnell: That's been Kevin O'Donnell: all right. So last So the question is, Kevin O'Donnell: the Kevin O'Donnell: the main thing. I I wanted to sort of emphasize in terms of that we messed up last time is that we just sort of drop the ball and took too long doing it. Kevin O'Donnell: So I think this time mainly. I just want to make sure. Kevin O'Donnell: mark you and I will carry it and make sure it's done anybody else. Just welcome the participants. So what I suggest is that I'll put together a little bit like James
Kevin O'Donnell: put together a mark down document or a Kevin O'Donnell: or a Google document. The discord and see if anybody has any Kevin O'Donnell: other inputs for it like a week before. So like on the twentieth Kevin O'Donnell: it's due Kevin O'Donnell: if anybody has it, and then we'll submit the Pr. Kevin O'Donnell: On through to the on the 27.
Kevin O'Donnell: Make sense Kevin O'Donnell: any other suggestions on what we could do better. There. Kevin O'Donnell: let's go on to the fun stuff. Kevin O'Donnell: Joseph. Kevin O'Donnell: you did. You've been doing some work on. So she's poet. And what's going on there? Why don't you give a summary of where you've got to
you, Chris. Joseph Livesey: Sorry I was on mute. Hi, everyone. And yeah, so I Joseph Livesey: this the past. Since we lost Matt, I've been looking into. So he's poet, mainly prioritizing the user issues that have been reported on the discord. Joseph Livesey: and Joseph Livesey: I've been The kind of outcome is a Pr that I've opened in the Docs repo which Joseph Livesey: tries to point the Joseph Livesey: the guide for the creating this sort of network guide, which is one that people seem to be having a lot of trouble with. Joseph Livesey: and it currently points to the Yaml file for sort of poet from main. So it's points it to nightly images. And I found Joseph Livesey: in my tinkering on sort of extensive going through that guide that I that it works. It's annoying to say it works on my machine right? But it it does work for me if I am pointing at the chime images. So I've opened a Pr suggesting that we point to
Joseph Livesey: the 1.2 Joseph Livesey: version of of of things that in that particular case, and otherwise, I have Joseph Livesey: run Joseph Livesey: our Btp, Ci checks on sort of poet, and they all passed no problem. Joseph Livesey: I've also used the It consensus the back end of our with one of our products and the that works fine as well. So
Joseph Livesey: and that's kind of the current stage. And then there's sort of a further, you know, the questions that we had about the algorithm and the build and stuff like that. That's a kind of longer term research project that I've got sort of on the backbone. Joseph Livesey: So that's where we are with that. Kevin O'Donnell: So are these C I. Bills on the Kevin O'Donnell: this off to me.
Kevin O'Donnell: that site, are they actually still failing with. Kevin O'Donnell: or they. Kevin O'Donnell: So it runs fine on ours. But it's gonna run on the main one. Kevin O'Donnell: Is it just need to be kicked again? And maybe everything's fine, or is it you don't? Joseph Livesey: That, I don't know. Andi Gunderson: Looks like the build on this Pr. I just need to approve and run them. Kevin O'Donnell: and they should kick off on the docs. One. No, I I think. Jose, I'm thinking of the actual software. I think the what kick this off was. We were looking at
Kevin O'Donnell: the Ci built on softly. Po, what? We're failing. But they look okay. Now Joseph Livesey: correct me if I'm wrong. But I think at the last meeting people pointed out that they they the Ci check, seems to have been working again. Kevin O'Donnell: it looks. The last valid build was June 2023. So this morning. Kevin O'Donnell: At least, it's running now.
Kevin O'Donnell: Okay? Good. So it seems to be kicking up well, from this it seems to be sort of Kevin O'Donnell: occasionally failing to test, but otherwise going well for the past month or 2 Kevin O'Donnell: next month. Kevin O'Donnell: All right. It's good then, me long running testing. So the this is currently it's got 2 parts. One is we're what I'm doing is basically putting together a
Kevin O'Donnell: good versatile helm chart for it. We actually already have it. But it's it depends upon some. You could keep custom stuff. So I'm pulling that out Kevin O'Donnell: and then putting in a a home test method so they can run. That'll make it so that essentially you can run it the test long running against any on any Kubernetes infrastructure, and it'll make it simple. And then we leave aside any of the terraform steps Kevin O'Donnell: make good progress on that. It's more about the the scripts that that we have to replace. There are
Kevin O'Donnell: fairly elaborate scripts for setting up Genesis and doing a lot of fairly cuts. And Kevin O'Donnell: it is only relevant to B, so that's that's what's got to be replaced. I might actually contribute Kevin O'Donnell: that basically simplified scripts that are are exclusive. Kevin O'Donnell: It's all, too, only stuff Kevin O'Donnell: do to help that out. But that's ongoing.
Kevin O'Donnell: The next item we did before we which we had before was a description of the tool chain that seems to have lost enthusiasm. However. one of the things that keeps coming up in the documentation and the questions from the users is actually the docker composed version Kevin O'Donnell: for 2 different reasons. One is that the Kevin O'Donnell: that there is actually a difference in the interpretation of the gamel specifically around scripts that happens between one and 2, Kevin O'Donnell: and then but all of our docker composes are actually in schema versions that can be theoretically can be interpreted by both versions of what I'm but what I've been saying recommending is actually the real problem we have here is making sure that Kevin O'Donnell: the all the docker proposed work on any doctor. that all the docker compose the animals to work on Kevin O'Donnell: either work only on the appropriate version of Docker. Compose, or if you can work on all versions of Kevin O'Donnell: the problem that shows up is that the doctor composed version 2 changes the name of the containers which throws basically the underscore turns into a dash Kevin O'Donnell: and that throws a wrench into a ton of the testing framework stuff on all of the Kevin O'Donnell: on all of the boxes. Core all basically every single thing. just because that run testing expects the names to be it a certain way, and it's formed it differently.
Kevin O'Donnell: so we have Kevin O'Donnell: 2 things I think we need to do. One is update the Dr. Compose schema versions where possible, to make sure that they run only in the right version of doctor. Compose, and then you throw an error Kevin O'Donnell: and longer term update those run test programs as they actually can handle Kevin O'Donnell: the new names. It's a Meyer, but it's a minor Kevin O'Donnell: technical change, but it's it's all over the place. So it's kind of irritating. Kevin O'Donnell: So what I think there is the right thing to do here transaction. As for volunteers, but to Kevin O'Donnell: open up an issue, summarizing exactly that Kevin O'Donnell: that makes sense. Kevin O'Donnell: Or does everybody understand what I'm saying there Kevin O'Donnell: And then the rest of the tool chain, I think, is in terms of documenting the rest of the tool that I personally think is less of an issue. It's it's primarily about the time compose
Kevin O'Donnell: any thoughts on that. Kevin O'Donnell: so James had volunteered to do scan the documentation for the raft mentioned of the raft consensus. Kevin O'Donnell: I think James is off on to other things right now. So it's unlikely he is gonna come back to it. But ken Mark, since you're already doing documentation, scanning for other reasons. Can you Kevin O'Donnell: take this on as well? Mark Carroll: I could do that. Yes. Kevin O'Donnell: So what? Remember what the goal here is, what we're trying to do is is deprecate Kevin O'Donnell: and eliminate the the saw tooth raft.
Kevin O'Donnell: Right? So we're we're looking for. It's just cataloging all the references to saw tooth raft and making sure that we either remove them or note that it's depicated. Better so that we could. Mark Carroll: Okay, I could do that. Kevin O'Donnell: Okay to Kevin O'Donnell: of my Kevin O'Donnell: picture.
Kevin O'Donnell: Okay? oh. I guess, Mark, you want to summarize what you've gotten to where you where you're going through with the documentation. Mark Carroll: Yeah, I've been kind of what? Mark Carroll: That was an initial question, I see which documentation to look at, the things based stuff in call or the other stuff and the docs repo. But I've been working through the 1.2 Mark Carroll: documentation of the Docs repo and Mark Carroll: but basically it largely in order of how it appears, and I haven't noticed any big issues beyond things that we've already mentioned. Like, the some of the poet stuff wasn't working for me, and there is a question about which versions of the to compose. I meant to be using to get this stuff working. But Mark Carroll: and I'm I'm currently working through the architecture section. I've never done that, too, and I'm not noticing yet Mark Carroll: any obvious, so big issues apart from the once you've mentioned that, I suspect the real test will come once I get far enough to actually try using the documentation to make some simple client.
Mark Carroll: That's the of I am so far, and I very happy to continue without this coming off. Kevin O'Donnell: I think it's it's right to focus on the sol to doxing. That was a good call way back when to move the docs out of the individual because the docs were causing Kevin O'Donnell: it depends. The update problems are causing problems in the bill for everything else. It's the pain Kevin O'Donnell: that's to keep it separate. So definitely, keep focusing on that. Mark Carroll: Okay, great Kevin O'Donnell: is the comments on that. Before we go on to the next one Mark Carroll: I think I should mention. My plan is, I'm Joseph. There's also noted for me. So I I want to know twice before I look at what? 2, 6. No, but separately, and in terms of what to do about them, my thoughts about to split them between things I can easily fix myself
Mark Carroll: and things that seem Mark Carroll: quite simple things. I can't fix myself. The I plan to create, to get to a few about. And then things that seem already of discussion that the time to raise in this code. Kevin O'Donnell: Yeah, that makes sense with them up and Kevin O'Donnell: different forms for different things. It's good. Kevin O'Donnell: Okay? Kevin O'Donnell: with that. Let's move on to Ryan Roberts, Palooza!
Kevin O'Donnell: Brian, this is around the Kevin O'Donnell: the first one was. We were saying. What the look of Kevin O'Donnell: look at what the impact of just moving to pros from the Protobuffs stuff in the South to the SDK rus. What would it have? What would it? What would happen. What would it break? Ryan Roberts: So yeah, so we've done that for our for our own stuff. Ryan Roberts: and it is purely a syntactic change. That's the the where we're using structures is equivalent. so it it's really going to depend on how people of how people have used Ryan Roberts: pro to buff in the application level code as to how difficult that might be. Ryan Roberts: if you use the declarative structural style to initialize things. It's pretty clean, it's cleaner than you you no longer need to use into. And things like that. If you've used, if you've done it in a more
Ryan Roberts: imperative way and use the set methods. Then it's going to be a bit more painful. But Ryan Roberts: There's How Even in that case, though it's it's a say it's a hands on keyboard kind of process. There's no real way for it to go wrong. Ryan Roberts: The type for equivalent and rest will make sure you don't get it wrong. it will ensure that fields are initialized when necessary, and things like that. Kevin O'Donnell: So if we made this change, we it would be, let's say, to 2 people who depend upon the SDK. If we made this change, it would be a breaking change. But or would it? Ryan Roberts: Yeah, it's certainly it's certainly a breaking change.
Ryan Roberts: but it is as it. It's it's in toxic breaking change largely. It's Ryan Roberts: a relatively easy one, a relatively easy one to deal with, and the types will. The types will lead you down the right path. It would be silent is what you're saying, right? Not not not not not silent. Basically. Just keep typing until the red squiggles go away. Kind of process. It shouldn't be, too. Ryan Roberts: should be too painful. Kevin O'Donnell: So if I remember right, the issue here is about maintenance, right? The existing protocol libraries are really maintained. Ryan Roberts: Yes, yes, yeah, they have a it's not. It's not officially marked it's not officially marked in Russia. It's it's applicable. But that call for Maintainer, and there's been no contributions for Ryan Roberts: for some time.
Ryan Roberts: and cross is what the Ryan Roberts: that it seems to me it seems to me that all the why, all the energy is app and it's maintained by the Tokyo people. Kevin O'Donnell: Okay? So Kevin O'Donnell: then, I guess the next thing to do is to actually just put together a Pr. Kevin O'Donnell: there! That justifies why we need to do it Kevin O'Donnell: right, just in terms of description of where it is. Let's just put the Pr in it, it will.
Kevin O'Donnell: And note that it will require a break and change the stuff, because that'll have impacts into everything. Basically. Ryan Roberts: well, at least we know it's not what it depends. If I can go about it. I think the I think the intens is to release Ryan Roberts: an alternative Senate source of SDK, which is Async. So if SDK Ryan Roberts: and then shim that into sort of SDK where we can. Kevin O'Donnell: I think we probably need to do both honestly. the just because they're it's kind of from the principle of change. One thing at a time.
Kevin O'Donnell: I I know the the so changing the pros because we're dealing with the dependency issues. It introduces the breaking change or the Kevin O'Donnell: SDK. But the doing doing the Tokyo stuff one. The Tokyo stuff isn't finished right? Kevin O'Donnell: Yeah.
Kevin O'Donnell: The. But the other is that it's a it's. It's biting off a lot more of the cake Kevin O'Donnell: you're committing to your at that point as opposed to just syntax day. Kevin O'Donnell: Bigger, bigger break is what I'm saying. So I think I still still think it's a. It's a good idea to do the pros change, particularly if it's just a syntax update, get the Pr together. Kevin O'Donnell: have it so that it runs through and and yeah passes checks. Then we can make a point of view. Eden. you have the Pr, they are open and mergable. It's not necessarily merged. And then
Kevin O'Donnell: at the same time, let's do the Tokyo stuff. Why don't you update on the Tokyo stuff as well? Ryan Roberts: yup. So Tokyo stuff, we have what looks like Ryan Roberts: a version, a an asynchronous Ryan Roberts: So it has low-level primitives, and it also has a high level of abstraction for Ryan Roberts: correlating block updates and Ryan Roberts: user and and custom source roof events. Ryan Roberts: so that's at the state where we should hopefully be releasing it. Publishing it before the Ryan Roberts: before before the next time we speak it needs for a bit more work on the on its unit testing. Ryan Roberts: We also have some enhancements to it over the capabilities of the existing. So for SDK, and then it can connect to multiple validators. And
Ryan Roberts: There's a Ryan Roberts: mechanism to be able to select which valid which validate which validate a node. You actually want to try to send transactions to Ryan Roberts: So that's a kind of a pluggable strategy pattern. But by default. It works on which one Ryan Roberts: currently has the highest block number. Ryan Roberts: so, and that definitely, that resolves a a bunch of issues with the synchronous source of Fdk, SDK, if you're accessing it from Ryan Roberts: asynchronous code which we've we've actually had endless issues with around shutting it down cleanly when it's running and when it's running and reactors, it's quite, quite painful. so I should hopefully resolve that and it should be a lot more lightweight. There's no Ryan Roberts: synchronization. There's no explicit synchronization required to in any of its any of its code, which is nice. and it all runs nicely without any additional threats. So it should be should be a lot more, a lot more likely to use Ryan Roberts: and to just to comment on the in the last meeting. we. We have completely removed any references to open Ssl, and as well it should make it a lot easier to compile to
Ryan Roberts: to awesome or anything else it doesn't support. Kevin O'Donnell: I'm curious. How are you able to remove the open? Ss, how do you do the signing, or is it just using for all the envelopes and stuff Ryan Roberts: I was? Yes, just using it. It's It's all just using K, 206 and Kevin O'Donnell: right? Ryan Roberts: So it's pure rust. pr. Kevin O'Donnell: So first up on that is getting that published. When do you think you can get that published out. Ryan Roberts: well, we have this. well, we can we? We can, as soon as we have the better testing for it.
Ryan Roberts: that should be publishable in the next next week or so. I can't imagine problem that Kevin O'Donnell: obviously so even the work that you're doing with getting this stuff working. It's it's probably better to get it sooner in the open rather than later, so the folks can look at it right, because the plan is to get it to open and see Kevin O'Donnell: what ideas can either be.
Kevin O'Donnell: Take it over directly, merge into the SDK, or incorporate it, basically open the discussion right? Kevin O'Donnell: Okay? The Kevin O'Donnell: So that's so like you, even the work you're doing right now. Better to switch it to do it in the open on the Repository publishes in there. We bring it up Kevin O'Donnell: next time. Basically do it. Kevin O'Donnell: Start, start a Kevin O'Donnell: start the discussion around the change because it's quite a. It's quite a huge change. It's a very good change, I think, but it's, you know, sort of people. Kevin O'Donnell: Yes, yeah.
Ryan Roberts: yeah, particularly the stuff around the our Our ledger abstraction is Ryan Roberts: possibly more controversial in the lower, that more more controversial than the lower level stuff, although it's not necessary to use. It was Ryan Roberts: It was a common pattern we we discovered in our in our integrations. Kevin O'Donnell: Right? That's that's a deal right, is you. You also. You never know what's controversial until you let the people look at it. That that's what. Kevin O'Donnell: okay, that's a review that was a bit long. I think we can
Kevin O'Donnell: close out some of the stuff and get more specific next time. Kevin O'Donnell: But let's any other questions of of stuff that I've maybe forgotten to include. I went through all the previous meeting notes. if not. Kevin O'Donnell: let's go on to the housekeeping.
Kevin O'Donnell: So just Kevin O'Donnell: what to do in terms of housekeeping. There are. Kevin O'Donnell: where you 2 main things is Ci jobs. I don't know of any jobs that need kicking from the Ci to get them to rerun Kevin O'Donnell: anybody. Disagree with that? Let me know. Kevin O'Donnell: or let the people know on discord and get it kicked the Kevin O'Donnell: the Kevin O'Donnell: that are open. I went through and listed out all the ones we have the Kevin O'Donnell: the Kevin O'Donnell: just going from bottom to top. There's the Java SDK updates. These are just depend about updates. they, they actually need to get looked at real quick to see why it's failing. Kevin O'Donnell: I will. I'm volunteer for that mark. I know you have a guy as well.
Kevin O'Donnell: Take a look at it there. It's actually the same. Depend about update in 2 different spots. It's for some of the test programs. Kevin O'Donnell: It's just that. It's just a the I remember with the defense. It's pretty pretty trivial stuff Kevin O'Donnell: the to to do, Joseph. Your Pr. 200 on soft suit docs, which is about the one to poet that still needs reviews. Kevin O'Donnell: Mark, you have a Pr. On SDK rust, which is this is the free after use. This looks like it's just a 5. Right? Mark Carroll: Yeah, it's just a security.
Mark Carroll: the Pd sessions. Kevin O'Donnell: the yeah. Kevin O'Donnell: Okay, it looks fairly uncontrollable. But that also needs a review.
Kevin O'Donnell: There are 3 in salt tooth core. What is the just based? Github build action which Kevin O'Donnell: I personally don't find controversial. Shawn actually put it through as I recall. Kevin O'Donnell: but that still needs review and put it Kevin O'Donnell: the another thing here. This is an Update of the salt to poet thing. I think this is actually, this is the syntax thing actually, for the I've got to compose Kevin O'Donnell: may actually throw in there. Kevin O'Donnell: But other, otherwise it doesn't look fundamentally particularly controversial.
Kevin O'Donnell: then the last one, I think, is maybe a little. Kevin O'Donnell: This is the arm, the build on arm dependencies. I don't believe they think this bills, but that's exactly because it's not doing the build on x 86. Kevin O'Donnell: The Kevin O'Donnell: I don't know. I don't think that this is. I've surprised that it only took this to actually change because we've tried to do this before. Kevin O'Donnell: I'm surprised that it took only these changes to get it to get the whole thing to build on arm.
Kevin O'Donnell: So there's a lot of other support tools that are. And they definitely have problems. Andi Gunderson: Yeah, I I believe the reason this Pr is still open is, Chris was planning to get back to it because it's not complete Andi Gunderson: like you said, there's still issues it. It builds. But I don't think it actually runs at like locally on arm. Kevin O'Donnell: Right? Kevin O'Donnell: So, Ryan, you were doing concert. Andi Gunderson: I I was just gonna like agree that I don't think this Pr should be merged. It should maybe be converted to draft if we can. Just so we don't lose the work. But
Kevin O'Donnell: let me do that. I don't have the Andi Gunderson: I probably can. Kevin O'Donnell: Yeah, maybe put it back. There, there you go. Kevin O'Donnell: But I was. Gonna say, Ryan, you did. You've also got a Pr one of our customers trying to do this. This arm build stuff Kevin O'Donnell: just for comparison.
Ryan Roberts: Yeah, I mean, it's it is as far as I understand it. Ryan Roberts: purely to do with Ryan Roberts: python python dependencies, and the them not being supported on that particular piece of One of the things I noticed when doing the overall Kevin O'Donnell: rebuild. The way that we do it is that there are certain there are certain python packages which actually are not published up on Pip, and they're only on Kevin O'Donnell: They're on the repo, soft with me as Debs. I don't see how to build. I I ended up actually. Kevin O'Donnell: but our building ended up reworking them.
Kevin O'Donnell: rather they are. I'm P. But there's no actual polished wheel for them to go Kevin O'Donnell: fiddling around. But what am I saying? The install installs A Deb that's being pulled from Repos off to me, and that Deb is not actually being built by the salt tooth core, nor any of the other SDK packages. And I think it's actually the sec, 256 stuff Kevin O'Donnell: for? Ryan Roberts: Yeah, that is that that that that that is an annoyance, because it is that the one that's actually got the problem Kevin O'Donnell: that isn't built for arm, or is it the pip level? Ryan Roberts: That is that that is an issue. Yes, yeah, on that. On that. I've been
Kevin O'Donnell: what would be better if we could actually eliminate the python Kevin O'Donnell: dependency for that bit of code, right? Because it's because if you look at it like the the salt tooth core doesn't actually have all that much Kevin O'Donnell: dependency outside of the SDK. Kevin O'Donnell: It might be worth going through and trying to eliminate those little bits of thought through this python salt to the SDK dependencies in favor of what's going on in salt. That's the K of R, Kevin O'Donnell: and then maybe this becomes easier. The arm, and also just the sort of Kevin O'Donnell: the tendency spaghetti that we've got. Ryan Roberts: I guess there's no reason I couldn't just be re-exploited from the rest. SDK and call from Python rather than Kevin O'Donnell: right well in general on the core, right? It's better to have a dependency on the the rust SDK than it is to have it on the python Kevin O'Donnell: in terms of just qualitative judgment.
Kevin O'Donnell: Let's go. That was maybe we'll throw in the link to your Kevin O'Donnell: Pr. That you've got open on our side, which is not a fit, obviously Kevin O'Donnell: not just to compare and contrast. Kevin O'Donnell: All right. I am now calling for So that's the housekeeping, and go through that sort of basically every meeting. I'll go through it. Kevin O'Donnell: do those list, or I don't know. We'll go through it as exhaustively first time. It's almost the most expensive. But now on to community care and feeding. Thank you. guys who have been doing Kevin O'Donnell: answering people's questions and stuff on discord. Everybody would agree. That's helpful, and it's always good.
Kevin O'Donnell: that, said I. not a fan of looking at it. But I personally, which is why they tend to go long in the tooth, but I went through this time and looked at all the issue counts and go through it, and what I'm looking for are volunteers and help Kevin O'Donnell: triage and see what's going on with those issues. Kevin O'Donnell: there's quite a lot of them Kevin O'Donnell: that are. If you take a look at them. Some of them are sort of plans. Ideas. Some of them are bugs Kevin O'Donnell: most of them are not. Kevin O'Donnell: quickly speaking about. Some of them are questions. Kevin O'Donnell: can I ask? Actually, I don't know if anybody wants to Kevin O'Donnell: the particular point for going through the issues. But can we? We all sort of make an effort to at least comment on them and get through
with the issues Kevin O'Donnell: or some chunk of the issues. so we can decide what to do with them that makes sense. Or is there a better way about that? Kevin O'Donnell: My guys. Kevin O'Donnell: this case, my guys didn't want my wonder. I am actually volunteering. You go through the issues and see what's what Joseph Livesey: understood.
Kevin O'Donnell: Request for the computer request from the community or orders for Btp. Kevin O'Donnell: But I like to go through those one of the things that we don't. I think the I haven't even looked at the the Kevin O'Donnell: the bugzilla that was on this for a while. I'm not sure we necessarily want to go back to the bugs in using it, because it was sort of
Kevin O'Donnell: that intentionally moved around moved away from a long time ago. So Kevin O'Donnell: I'm thinking, why don't we just focus on the issues? In which case, then, don't we need a set of templates, and with our sorry labels it's pretty awesome. Kevin O'Donnell: And does anybody know a good example of them? I know, like the Kubernetes core, and things like that, for instance, use labels to triage issues and basically just Kevin O'Donnell: are pretty native onto Github. Kevin O'Donnell: Anybody have any thoughts about that? Kevin O'Donnell: We haven't to date done that. Any kind of labeling like that on
Kevin O'Donnell: All through stuff. Andi Gunderson: I am pro labeling Kevin O'Donnell: or labeling. Yeah, you know what I'm saying. Like, like, I have a label for triage. I just think somebody needs to go look through this. And this is cool. This is dead. This is fail. That kind of thing. Kevin O'Donnell: Okay. Kevin O'Donnell: did. Kevin O'Donnell: Well, there's a built in once. I wait. I'll Kevin O'Donnell: if people can look around for a set of labels on this. There, there's a standard Github labels, but I don't think those are sufficient. But let's look around for a list of labels. We'll b around on discord what they should be.
Kevin O'Donnell: and then you don't think I can. I think they're controlled on Github, as I recall, so I don't think I could set them up so, Andy. Kevin O'Donnell: so you could help us out with that and just start using Kevin O'Donnell: starting on issues. I don't know so much on Pr, because we got controlling build on Kevin O'Donnell: labels. Andi Gunderson: Yeah, just a note for people who are going to be looking at issues. If I,
Andi Gunderson: Peter or Shawn, added it. They're probably just stories that we had in the backlog on Jira. Andi Gunderson: that we moved to different repos because we stopped using chero. So I did remember that. Okay. Kevin O'Donnell: so that's what I thought it happened.
Kevin O'Donnell: Okay? And then that's and then those stories become things that people can pick up them or evaluate whether they're still out. And if they are like, people can pick them up right? Andi Gunderson: Yep. Kevin O'Donnell: no. Kevin O'Donnell: So Kevin O'Donnell: any other ideas, contributions, things that people want to throw out there Kevin O'Donnell: once, going twice and to open discussion. If anybody wants to talk about anything else they want to do. Kevin O'Donnell: The last is Kevin O'Donnell: anybody like me to run this meeting differently. I'm happy to take. Yes, James Kevin O'Donnell: also. Would anybody else like to run the meeting?
Kevin O'Donnell: Okay. Kevin O'Donnell: any other is Kevin O'Donnell: but once going twice, I will Kevin O'Donnell: update with my notes and public afterwards, and let everybody anybody can comment on the confidence about what's going on. If they disagree about what's happened Kevin O'Donnell: but once, going twice, 3 times. Kevin O'Donnell: Alright, thank you. Guys. Joseph Livesey: Bye, everyone.
2023-07-04 10:13
