 Welcome to the microp 21st of September, Dead Sink. And today marks the beginning of a new sprint. So let's go over our last sprint and check out the status of that. And then we'll have a discussion about the new milestones that we've put together and how that's going to affect the next few sprints that we want to plan out here. So, Chris, do you want to take over and show us the current sprint? Okay. I think it could be argued that the sprint was overplanned. There's a lot that didn't get done, but go ahead and go through it. Let's keep it high level except it's like. See, so let's try to keep it high level if there's stuff that's not okay. Not even started, then let's take a look at that and try to break it into chunks or the epics that they're assigned to find a pattern to what did not. It seems like there's a lot of stuff in review and a lot of stuff in progress. Yes, so I'll go ahead and get started. I did finish the code to upload the precise data in the Selene is merged into devs so that is done. I also have a pull request for storing the liquid samples and the data deletion stuff. So that those three things will be the next version of Selene. Once those get reviewed sometime this week, I will do a release of Selene that includes those things. And then I'm working on converting the existing Maria DV for precise into Selene. I had the data from Ken and I'm just working through that. And this is all really do a quick word collection and tagging and then I'm, you know, I did not get to any of the, any of the tagger stuff last week. So, so yeah, that's also got the, you got the WordPress drop done, which has been lingering for quite some time so that was nice. So yeah, that is my update for the sprint any questions on that. Yeah, they're all the next step they're doing the tagging UI and the end of the backend stuff that needs to be done for that. Your stuff is either in the to do category in the interview category so you should be able to get on the same on the next issue next to the issue. Oh, yeah, yeah, I certainly have plenty to do. I have one in progress there that I'm still almost done with the conversion. And then I'll move on to some tagging stuff. Like 10s issues. Oh, he's talking. He's just. Oh, you're anybody. I couldn't see if you should move or not. How do you know what I'm lying my lips are moving. The expand vocab of yes no the low hanging fruit piece is done the ticket was upgraded with the extra. Yes knows that I put in there but I don't even know if it's worth releasing. It's just, I don't think that's the issue. So the next issue for issue is the line turnaround stuff. So, you know, that's not in progress yet. It's going to take some work. So I've left that in the to do category. Same with the message bus visualizer on new model creation. This is kind of waiting for the process. So that's why it's in progress. All the codes ready to go. I'm just waiting for the database to be turned around and then I'll convert the select from the old Maria dv structure to the new structure and then we should be able to integrate. Probably be a couple of days of work on that. No audio output on first boot up. To the best of my knowledge the sg 201 fixed that. So I'm leaving it open for now, but I could theoretically close it, I suppose. I'm currently actively working on the next ticket, which is the sj 201 based enclosure skill. We look automate precise model validation tool actually new model creation is done I created a new model. The flow is 69 because I already created that and created the branch repository that holds the default model that sent out to everybody and the new model I created with instructions on how to update your model. And that was done as part of the project for. I keep calling it project overload but for that project. So we can yeah we can say that's done. That's that's fine. Over here automate so automate precise model validation tool is kind of what I'm working on. We go back for seconds. Yeah, so with the new model, do you think that's something that we should, you know, push out and invite people to test out properly. You know, okay told me that it didn't seem to work for him. He gave me a thumbs up. Okay gave me a thumbs down. So I was on the side and waiting for feedback from everybody else. He said that it didn't improve the experience for his significant other. And that was kind of the goal. Well, I'd love to believe guess you can leave new model creation. Yeah, I think I think very easily both could be true. More feedback but my point being from a ticket perspective it's done. The new model is up and available to anybody that wants to pull it down. I gave it to Jarbas and he promised me he'd test it but he hasn't yet the last time I pinged him. He said well I've been kind of busy and well I just wonder if we do a we do a forum post that says, you know, here's a here's a new model that, you know, can's made like if you've been having trouble with the other one feel free to give this one to go. Here's how let us know what you think. Maybe just could assign that to you. You could do that and then close it. Yeah, because you know, if people are already having not a great experience then there's no humming in them trying something else. Right and the automate precise model validation tool is what I was getting at to me that's the end result of what we're working on as new models are created. So that compares after it runs them against default test set and see and looks to see if the recognition is better than the prior model. So that's going to happen automatically. Once the database is ready. And so that's what I was getting at. That's the ticket that once I integrate a couple of days and change the data model for the selects. Chris you and I can talk about it. I don't know if it's going to be selects or if it's going to be an API. And I can go ahead and put this one to rest, but it's kind of blocked right now. So that's why it's in progress. And I haven't started on refactor intent service yet. That promises to be fun. Yeah. Can you open that one. The intense service. Okay, so is this one that is necessary for the current milestone that we're working towards. Probably just because it's a refactoring issue. It's, it's, I mean, it doesn't, I mean, it's, it is a literal refactor so it doesn't change anything. I've had a look at the code and it looks like a good. It looks like good changes in terms of this isn't worth to do this is reviewing some. This is just just a PR review. Yeah. It's a full request review. Yeah. Can I assign to it to review the request not to do any coding. Okay. My mistake. I'll look at this and I'll do a code review. I'll move it to review. I think I was a little confusing too. Yeah, I'll do that for a second in the morning. Sorry, I didn't realize that. Yeah, I'll get to that first thing in the morning. The, it looks like core three 24 the vocab one is either done or should be in review or in tests or should be renamed because it looks like you're interpreting it as solve the roll over issue. But this was one of the many ways we could have solved that issue. And so the idea was to try this one, see if it solves their problem for them. And, and if it does then we, you know, then we can close it out if it doesn't we can still close it out and just solve the problem. The other thing I meant to document in I didn't. And I'm pretty sure by default, the config setting that says display an audio alert when listening. That's configured off in our default configuration. Is that a correct statement? Sorry, I miss that. Say it again. We have a configuration setting that says play an audio alert when we're going into listening. And I believe it is turned off by default. Is that true? Yeah, that's true. I think on your device on the mark to protect cities. So my point being that if we were to add this could be an education issue. That if we educated them to turn that on and wait until they heard that beep. There yes no problem would probably go away. Yep. Okay. So we recommend we do there. Update core 311 to include that as a possible answer. And send them that info. See if that, see if that solves the problem for them. Okay. Okay. The work to expand the vocab is this over this work is done. Yeah, it's done. Ken says it's not convinced. It's not, it's not in the it's it hasn't been merged at all. Right. I haven't seen a pair for it. I don't even know if it's worth going through that process is my point. I suspect the educational approach would bear fruit. Well, I think, because I think other people have had issues with the vocab of it. So I think this is one of those, you know, different circumstances for different people. So I think integration tests for this or unit tests. Right. A lot of the audio stuff is really tough to test. I was going to ask the same thing about the intent service refactoring. If we had spawned any VK tests, do you catch capture that? I think the any of any test to test skills automatically tests that because it's intent based. So as long as the skills are still working, we intend to still working and the refactor work. I think would be the idea there. Yeah, but you're right about the, you know, audio for yes and no. We need a whole. Thank you. Something it's a bridge we're going to have to cross at some point. It's not clear to me where our VK tests run, but if they run in the cloud and a headless server, then that's not going to work because headless server doesn't automatically include devices for non-existent hardware. It's on the roadmap to expand the VK tests to start, basically expand the pipeline right now. It's you inject a bit of text that goes into the intent parser, you know, and then out from there. But it's on the roadmap to be able to start a test with a sample bit of audio coming in, you know, probably with specified timing between the next audio sample and things like that. Right. So that's on the roadmap. So what should we do with 324? That was my question to Michael. What do you want to do? I'm inclined to run, just run the regression tests on to make sure it works and then merge it in. 324 is the yes no. I think, I think what makes sense is I'll update 324. I'll put in the informational approach. Who's going to actually interface with our contacts at that place and tell them that. Because that's who I would want to assign it to. In other words, the last comment will be, here's what you should tell them. And then here, assign it to the person that's going to tell them. Do I hear any volunteers? Josh is in here. I've been more the point of contact on the technical side. So, um, you know, if you've got any of them a new image, because we can just turn the sound back on in the new image. If we're doing that. Well, it seems a bit heavy just to have them go in and change a config parameter from false to true. Um, but I can tell them, I can give, you know, I can tell them the, the one-liner to turn that back on. Or I can give it to Derek or whoever's going to meet with them. Yeah, we. I think if it was worth setting up like a shared drive with them, I think. We couldn't, you know, probably do that. Add some extra documentation and stuff in there as well. I'd be useful. Okay. Well, why don't we just have guys email them the solution? Right. So I think what I'm hearing is assign this to guys. And be done with it. Yeah. I think we need to spawn a new ticket though that addresses the, the bigger issue that you've found around. I was just creating that. And possibly one that is a. That's a really important issue. I don't want to go into that yet. The next, the next, the next message tip of the iceberg, which is. If there are people that are having trouble having their. Yes, no, in their language being recognized. Because. That really becomes a backend issue. And I don't want to go there in this meeting about what we would have to do to fix that, speech to text. And the fact that this is then going to also have to carry that language information up with it. And I don't know like, you know, for me yet. I don't know how well the recognizer is going to handle me. But anyway, my point is there's actually two sub there's two different tickets here potentially. The one that I mentioned, which is line turnaround, and that requires refactoring and speaks to margin. And the one that gets talked about, which is the community complaining that they have problems having their languages yes, no being recognized. Well, it's more that they're doing it in English, but then, you know, English may not be their first language. And so with accents and stuff, then things don't always get interpreted the same way. Is the yes, no question. Is that does that go through the translation services or not? Yes. Oh, well, as in, if someone has their language set to German, it will look in the German directories for that directories for what yes and no is in German. Yes. Yeah, but what I was getting at guess is when it finally makes it to STT in the cloud, is that information relate along with it? So it goes to STT first, it turns the transcript of the STT and then Microsoft checks it's Microsoft checks the vocab within my core or the skill for whether yes, no, just or what yes, no actually is in that in the language that I've said it to I'd recommend we have a brief discussion after we terminate the meetings recording to like and expound upon where I'm coming from. Yeah, I don't want to get too much into problem solving on these problems for sure. So, right, definitely take a note of the sounds like the two tickets that we need to create this and let's move on. OK, so that's the status for Ken. Do we sorry, I just thought one final thing, the reducing the turnaround might turn around time. Do we want that in the next sprint or is that going to the backlog? Go ahead and put it in the next sprint and then we'll decide. You know the creation. Yeah, so I feel like. Somehow, you know, looking at the board makes you realize you need to spend more time putting things on the board. It doesn't feel like an accurate reflection of the last two weeks. But, you know. So I think like there was a number of CIS fixes that. Did get through all the way to the end. A lot of the stuff that I ended up diverging to was documenting community processes, which is so that M.I.C. 461 in review, which is the. Essentially the handling PRs when they first get in there. And then moving to the bottom one in in progress, which is around how we can tie in feature requests in a more. Structured way. So there, I think. Quite important and. Quite useful to keep going on. Posted the PR labeling one in in the community for feedback last night. But I don't know that it'll get that much. Interaction. Just because it's very developer focused. The spec out media playback system is a is a big thing. And I think it needs to happen for us to be able to move forward. With that. It's kind of stored a little bit just because I haven't put any work into it in the last few days. Because I've been focusing more on the PR stuff. And those the two bug. The bugs in the in progress column. They're still hanging over from the. The service status stuff. And then a lot of the stuff down in the. To do column is all requests. Except for the bottom couple, which are. Yeah, things things that are. Semi important, but not urgent. And so I bumped them down the priority list. And then the posting them out to images for the community. I. Realize that. It needed a lot more context than just sort of. Dropping the two system images out there. So I started writing a blog post for it. But it also just felt like less of a priority than getting our. For request system. Flowing better. So that's where things are at at the moment. And. So you're primarily focused on like getting the community interaction process. Yeah. Yeah. Okay. So yeah, I still got a lot of stuff in progress. A couple of these things are actually. More like Kevin's stuff, but. Yeah, so like an in progress, I've got, we've got evaluate PCBA for. Simple as J2O on. A lot of that is actually. Stuff that Kevin's been tackling, quoting. The next round of prototypes, but. There's still some things we're working on. So. They're, you know, that he's addressing in terms of. And kind of talk about on Friday. So not quite ready for that. Yeah. And that's the next ticket to SJ two or three display driver. I think we could probably. Back burner this or put it as not. Won't do maybe this is the. We're thinking about creating our own display driver. But at the time, at this time, it's not really feasible. Because it's the complexity. So we're looking to just source the display with the display driver included. So the next one is I was working on that a bit on Friday and today. That's the first FDM 3D print of the SJ to a one based design required. FDM variants, the SJ to 40. So yeah, I've been working on that. And I've got one part started on the printer for that, but it's likely going to take me. I'll just even printing it takes, you know, half a week just to get everything printed. So it's going to be an ongoing project this week. And the two gooey. How long does it take to print? Like really, it takes like 100 hours. Well, whenever I first print something the first time around, it's like, I don't have things really like printed as a kid as it gets done, you know, so like, I don't optimize and nest things in multiple parts and stuff. So like, oh, I got his part done. I threw it over, you know, and then it's like 12 hours. So yeah, it takes a while to get to a point where like, I'm doing four or five parts in one run. That's just kind of how it is. But yeah, yeah, eventually, you know, with the decent size build platform, you know, nesting like some optimization. Yeah, like be able to get it done, you know, one day. Yeah. So, and then the last thing there on the progress was the GUI for the tagger. I kind of worked on that up until about Thursday last week. And then kind of still some several things need to be completed there. But kind of got busy on more of the part two stuff and updating slide deck. So I had, I think it a chance to complete that. So partially because of our discussion on Friday, two of our priorities and Chris got distracted with a number of things as well last week. So I don't think he got, I hope I was blocking him. I'm not having that completed. No, you weren't blocking me. So you might be. Well, I think it's, I think, yeah, I think right now. I can dedicate. I can get that done next there too. So yeah, like the updated slide deck that's in the done column. We've got prototypes. Ken, I've got Chris's. Let's find a time tomorrow. I can drive this up to you. Maybe you can meet halfway. Yeah. Sitting right here on my desk ready to go. Yeah. And then the other thing that I don't have on here, but since we talked about it is, is updating some of this bomb stuff. Michael asked Kevin and I to kind of tidy up the bomb and get it a little bit more ready for a realistic pulling. So Kevin's kind of approaching it and his end with the electrical engineering stuff. And I've got all the kind of other things like the display with bigger drivers and screws and cables and then, you know, the plastics as well. So getting all that organized and ready to, to quote for not only just for the depth, the first round of depth kits that don't require housing, but they will require all the parts, the speakers and the cables. But then, you know, getting ready for the big time as well. So I started, I started working on it a little bit. Yeah. So that's kind of what we're got in the spread. I did throw a couple of tickets in the backlog to make a prototype for you, Michael. So be ready for, for when we get a board that you can have and that one for guys as well. This is going to be the laser cut. It's on. Sure. And then Josh has actually already gotten the parts cut for his side shirt and parts. He went ahead and cut himself a set. Yeah. So yeah, that's where, that's where I'm at. I still think a lot of this is going to take me at least a, probably a week to finish. One thing I've been thinking about is how far to go. Like the bomb itself is one aspect of it, but going through and getting more realistic production quotes and like a whole package ready, you know, quoting out the plastics, like all that can be pretty intensive in terms of, you know, sitting all that stuff out and following up and everything. We're not ready to sign in the contracts about, you know, milling out an injection mold or anything like that. So we don't do that far down. Okay. Yeah. Yeah. At this point, it would just be preliminary. I'm going to give us some bulk work stuff. But yeah, we can wait. So I might need to load a couple of things in the backlog, but one of the other things too is this camera evaluation module bit. I haven't really done anything on that yet. So I need to get, I do have one of the old first gen cameras. And you know, at least plug it in and see what the quality looks like and stuff. We should add another ticket for actually choosing a display module now. Since we're not designing a display module, we're going to find an off-the-shelf one. Yeah. Yeah. Oh, yeah. Okay. Yeah, I actually do need to do this. So I have been chasing parts to display and for this, the speaker drivers, they're speaker drivers like way too expensive right now. So I can add tickets for that. Yeah. Okay. Yeah. Thanks. Okay. The only people left are Josh and Michael. Do you know where. Josh is on the last pass credential stuff. I knew he's going to clean it up. I haven't heard anything. I will be. Is. Audio popping issue. I'm actually going to meet with Kevin later today. And sort that out. I don't know if there's a couple of. Possible solutions, but we don't, we don't have a good handle on it. Okay. And. Limited revenue planning. Yeah. Actually Johnny and Chris. So. Should we complete this sprint and plan the next? Yeah, we should. We can. Yeah. So yeah, just in the, in the retrospective sense. We seem to have an awful lot. In there that didn't get, we never didn't get to four or another. So. Put anything based on and so it wasn't necessarily. So I guess that's a question for you guys. Would you rather. Build in some buffer time. Knowing that we're going to have interruptions. And. Then the only downside to that might be that. You complete your work early and. You have a couple extra days left in the sprint to start working on the next one. Or, you know, plan to try to fill out the two weeks and, you know, expect that we probably won't get to everything. I would rather have too much to do personally, but. I also, you know, I don't want to give you the impression that I'm not doing what I'm supposed to be doing. So. I almost agree with Chris and that. I don't know. I don't know. Building in buffer time at the engineer level seems a bit awkward. That's usually a managerial device. All right. I just wanted to give you the option. I mean, I did just watch. Lower decks episode about buffer time. Either way, there's always more to do. It's just like, right. Yeah. We have a walker of a sprint plan right now. Yeah. Let's go through and see if we can. Do we have a, the following sprint? So this one's going to be 15, right? Yeah. I just closed it and everything. 15 already. And there were a few issues in there now. Everything from 14. It wasn't done. Yeah. So I think that we should create a 16 sprint and go through and try to. 16 is created. Okay. So let's, let's go through this and try to push things into 16 that are not. You know, take a look at this from the point of view of our first milestone, which is to get dev kits to all of our internal developers, right? We want to get Mark choose to everyone. So if it's not serving that purpose and it's not serving the or the next milestone, which is getting dev kits out to the developers in our community, then, you know, we should see if there's, if we can deprioritize those tasks, you know, and put more focus on the things that will get us closer to completing this first two milestones. I would say just the first milestone, but, you know, in particular for this one, I know that there's a lot of blockers that are sort of out of most of our hands. With respect to the hardware. All right, I guess we'll start at the top. Upgrading to Angular 10 is probably the first thing I'm going to do before I start adding logic to Salini. So for the tagger, you can get the upgrades working. So I think I'll leave that in there. The database schema, basically what I've been doing is implementing what we've talked about piecemeal as I need it for the different pieces. So there's some still some tagging tables. I haven't implemented it yet because there's not any code that uses them. So that's what that is. Dependencies for different tag types that came up out of the discussion. We had recently how it was possible for one tag to say, okay, I'm not going to need to tag gender if it's not a wake word in the first place, that kind of mechanism. I think you probably need to talk about how to implement that. I don't know if I might put that in the next sprint. I don't know if we're going to get that far in this sprint. Okay. Tagging UI usage metrics can't be done until a tagger UI is done. So I'm going to move that as well. The mark to give you no audio output on the first boot up. Let's keep that in there. Okay. Keep this one. Okay. This is in progress. These are in progress. Okay. This one is still to do. This I think involves. This is also a little bit related to. The phase zero or milestone zero because we don't really need to get the existing image up to 2008. So I guess my question to you would be, should I do this? First, and then move on to the tagger stuff so we can. Move on to the tagger stuff. Okay. I don't understand what the benefit of upgrading. Just a little, we'll stop asking you if you want to upgrade. No. Okay. That's a quality. Which is. You could look at the ticket that was. Cut for rollover and set and add it to your blacklist. And it'll also go. If you just add that skill to your blacklist. And your config file. Yeah, that was. The other alternative was just turning off the update stuff for the roll over folks. Instead. Yeah. We don't have a ticket for that. We'll definitely push this one. Okay. We had a ticket for that. It was closed because that. We had a ticket for that. We don't have a ticket for that. We'll definitely push this one. We don't need that. Because that was the solution. And. I don't think we ever. Did the research to figure out which skill. Or where in the. No, I have. Okay. It's in the ticket, but I don't know who. To, you know, when it comes to this point in these tickets where it's like, you know, I'm not sure who to assign them to is the issue. Perhaps we should just assign that to Derek. Which, which ticket. This is communicating to project roll over that. They can turn off the update by blacklisting the skill in the config. Which I can't say the comments. Which ticket was it in? Oh, I thought I put it down in 16. Hold on a second. I thought you said it was sold or bought it and it goes. And so. Thank you. Okay. No. We'll talk about that. Take it we're talking about was assigned to me. So theoretically it should still be assigned to me. So it might not. Yeah, let me. That's what I'm saying. Let me look at. I have a little filter here. Yeah, but you're going to need to know which ticket it is. Right. Yeah, I know it's Mark two. I just put it in sprint 15, so. Okay, okay. Okay. First limited run of Mark to dev kits, I assume that's still we don't we'll get past that. Adding a firewall to give the image. That requires the new image to be built. So I think I'm putting with the other one. This was still. Yeah, we don't. That's still open until we've given them a solution. We should leave that one. All right. Are we going to do 20.8.1. In the sprint. Yes. Okay. 20.8.1. Yeah. Yeah. Yes, media type support is so is that related to the whole, the larger issue of how to do. Media is this, are we putting this off? Are we just going to know what he's got so far and take another stab? No, I think, I think that one needs to. Needs the. The service mapped out first. Okay, so we can take this out of 20.8.1. You think. It depends when we release 20.8.1. As in I haven't got a, I haven't got a ticket for. You know, do the 20.8.1 release. And it's just things that are. Contributing towards it. Okay. Sorry, leave this in the sprint for now. I think. Most likely bump it. And then if I, if we magically get to it, then that's a great, but. Okay. This is labeled as 20.8.0. Yeah. Yeah, it leaves that one in. It's just small change. Yeah, just small change. It's just small change. Um, Remaining things that no longer. Did on it anymore. But this would be one right? Yeah. Yeah. We don't have a doctor. Oh never mind. That's the skill thing. So it's 20.8. Um, I don't think. This really speaks to any of our media goals. No, that should be moved out. in the different epoch we have a security for the security changes you're going to make I think we've got too many epochs at this point maybe but user experience okay and next brand or top of the backlog okay verifying and install is this I think this takes the larger issue that we're investigating regarding updates in general this job is concerned that an update fails how does it restore back to its previous checkpoint so I don't know yeah this should go into the milestone three sorry milestone although we've got to resolve whether we're going to do uh make this a requirement for the developer versus the consumer release let's put it in the consumer release update our milestone at the very least because it's got to be there right so put that into the college micro for commercial loads although it's technically not really a college issue we'll sort that out okay um we probably will get to that in the next print yeah put on the top of the backlog okay no this is in progress we'll leave it there should finish it this week you know this has got a couple sub tasks in it you got the model creation done but tagging and data capture data capture are still being worked on so we can leave that there upgrade matter most so I think a good pace for me probably is to upgrade do something like this once per sprint um we did the um the public website last sprint um guys you think it's more critical to get up matter most upgraded or discourse well actually discourse maybe does that'll unblock one of my things okay so I'll move matter most to the next sprint and let's make sure there is a you're going to create a ticket for discourse upgrade no I have no okay you want to create that I hope I can put it in this sprint yeah in this sprint oh maybe I did I'll find it and put it in either way okay this needs to be reviewed review some messages pull from the last sprint in progress uh skill api I think that's probably needs to go to the top of the backlog I know there's a there's a PR I think there's a PR out for this right it's an open PR and there's documentation um but we haven't really is this one of the things we wanted to take a higher level look at like a specs back it out that kind of thing because it's well hmm I'm trying to remember what the the bluffer was let's update the description then and um come back to it next time or leave it in if you want I'll take a look at it you know I mean I think it's good to merge I think yeah yeah probably it just needs needs another review that's all my request to changes they've been made okay same with the plugin system uh yeah I think it's from memory it's good to go I love this was coded up oh it's in a pull request yeah in review guys you can get to this in this sprint um I mean we need to do the download I've got do the mark to September update um and so we're not really going to do two blog posts um in quick succession like that say I think bump it okay so this is a bug this is it should we put this in the user experience epic um is there any comments standalone or I think there's the I think there's tickets listed down below okay yeah yeah okay and you said you were still working on these were dependent on the status things so we'll leave those there yeah yeah that one's um there's we're leaving Derek's stuff in there that's in progress Derek Derek so that message bus visualizer task I mean that's that may be something that's really useful especially when we get into the discussion of the timing thing that the 10 was talking about um but on its own I don't think that it's basically it's a tool to help with other issues so um yeah it can move it can be pushed let's let's just push that into the backlog why do you data file names and their hash value I think this is a discussion needs to happen this sounds like a way to tag your issue too right yeah okay you and I can just talk offline about that the ab testing I mean we're not going to try to get that through that the sprint right we're not down there yet hold on those sound like they can be pushed audio region tagging what this is the whole like extracting one of the way for examples out of so okay whether we're going to tag regions or extract them and make them their own things you know we need to make a decision about that can that not go to the next sprint uh depends on if I get to the point the gooey or not the architecture of the I mean in a yeah that would be why so yeah I'm not push gooey b at the very least down to a future sprint we're not going to get two of them done and then that means that we can push ab testing to the next one oh I'm not giving Kevin access to Jira that's my bad we'll do that bomb review uh I bumped that one Josh is doing this I've also bumped that one even though yeah that's just a pair okay yeah it's in review uh yeah this is when I threw in there just because I was seeing a lot of people complain about wake word more than usual so I wanted to kind of dig into it but um I think I'll post the instructions for Kevin for Ken's model first and and just sort of invite people to holler at me if they're having issues and we can try and dig into it a little bit a lot of the wake word stuff that I'm seeing on the forums is because they're porting us to really tight environments like pine phones and things like this and then saying well precise it's consuming too much cp let me switch over to pocket spanks and then it doesn't seem to work that great and then that's where you're seeing a lot of the chatter from yeah yeah yeah if people if people are complaining that pocket spanks doesn't work well then that's a very well known very well known thing not our documented feature we didn't write that one yeah well and it's in the documentation like it's entirely configurable but it's entirely not great so all right fine what is going on with Travis yeah I haven't seen heard about this yet uh it just it was so the the the build result was just being stuck at you know pending um for some of the tickets there was a bug an outage reported in Travis before I saw this that was the same behavior and I reached out to Travis and they basically just got me to refresh my credentials in their system and you know which shouldn't really affect this but I haven't seen the issue again yet so do we not host our own Travis server no we host our own Jenkins so then why are you from a high level what's what's the purpose of using Travis instead of Jenkins we use multiple for different purposes I think eventually we'd like to get to one but right now Travis is doing some things with cores pretty much specific to core isn't it like yeah that's core again specific versions of python and some other stuff um but we just haven't gotten around to it yet yeah and I think it's just because Travis was set up a long time ago and it's so easy to get started um on things on certain things so um yeah particularly that like building building packages across different versions of python and that sort of thing so um yeah I guess where I was coming from is is I mean if we host our own Jenkins we know that's free the only cost to us is the cost for the servers that we house it on is that the same model for for for Travis or will Travis Travis is free to open source companies yeah so Travis is managed like we're not hosting it but I don't believe regardless how big we grow it it'll always be free as long as you say open source it just seems like we shouldn't have both okay we shouldn't and we have a plan to get off of both but one of one of them but that's just lower on the priority list right now um so I think leave that one open and I'll I'll just keep an eye on whether it still exists and this was the um the issue where micro skills were still building against the 20 to core image and so okay uh did a sort of quick fix by modifying the Jenkins file in the micro skills repo um but we I think previously there was a void conf builder there was a particular um docker image on Hudson um that we were using instead and so okay Jenkins file just kind of like mimics that as opposed to using the proper docker image yeah so we need to revert his and and create the next create the real docker image from core to use right yeah yeah that shouldn't be too terribly difficult um Hudson Jenkins Travis sorry Hudson is Jenkins is one of that Jenkins instances yeah we have two Jenkins instances we call one Hudson and one Jenkins to try to avoid confusion Hudson is the one we're trying to get everything on but we haven't yet all right um so uh this sounds like there was a uh an error in the uh deployment process is that right this was a an error that came up from the 2008 release something we need to add to our 2008 checklist right uh basically we need a new version of the void conf docker image for each major release okay so I'm not gonna add it to the checklist somewhere right yeah okay Chris did you want to do this in a way that sounded like you were typing yeah which is totally okay but I think it's done sure it's done Chris did you want to do this you want me to do it uh if you're if you know if you think it's going to be easy then I I think go for it tie in to the system voting process so this is um yeah creating that process for the community to request features and and vote on them and and all that sort of stuff okay so here's the discourse ticket that's the infrastructure ticket I'll do this sprint well if one of the objectives is to kind of get some of these out of this sprint because it's too big are we really going to get into tie-in feature requests and voting processes sprint well Gez is more community focused than uh product focused like specifically uh so Gez is sort of running in parallel rest of our milestones I actually think we could probably bump that one and move in the tidy up road map for initial release um so maybe bump that and I'm gonna move a different one in there in its place okay which I think should happen first anyway um so I want to take all the work that we did around the road map stuff you know looking at what's on the road map for different technologies and different things and get that um get that into a format that we can share out and you know get input on from the community um and so yeah that will either be through Google like shared Google document as the sort of master roadmap um but then that links out to individual you know technology or project-based road maps um which could either be a google doc or potentially a project board in github um yeah okay all right and the mark 2 september update is that a blog post is going out this sprint what's what's our grade this course this course is our um forum um um community.microft.ai it's uh it hasn't been updated in a while so as I mentioned before I'm gonna do I'm gonna try to do one um DevOps type thing in each sprint so we don't lose track of these things um last one should be a button click upgrade presuming nothing goes wrong yeah but I'll need to do it in the evening so that we don't I think we have a like an eight o'clock p.m. eastern um time that we do these kind of this type of work so um so yeah do we have backups of the system is this a system that we yeah yeah of course the matter of course they should they both have backups uh september mark two update is this thing you're writing the sprint yeah it's a thing that needs needs to be written okay does anyone else want to take it on fueler well it's a mess that's for me anyway all right so I can help you with it all right assign it to me you know I'll go with people to get it done I think you can push 331 into the next sprint Chris okay so now what we we've been through that god there's so much stuff in here um a lot some of this stuff is already in review and stuff so I think that's probably a little looks bigger than it really is um so one thing we haven't done then is go through and what types of things do we need for milestone zero to get done this sprint right right uh so for something I just noticed that I was looking was the laser cut housings for gezen michael's that something you're going to get to this sprint um is that that's part of milestones zero right um yeah I mean it's well I do not things are still kind of influx on um whether um you know there's going to be any boards to go along with those but but there's no reason I can't get the housings done and set out and be waiting the boards can just be sent directly to michael I guess independently so yes um I'm going to try I'm going to do that feel free to hold sending me stuff until it's complete if you want like well if it's easy to send it separately then that's great but yeah I mean I I guess it's not I mean I don't know it depends it's good all it all comes has to go through Kevin before it gets to me so but okay yeah I can I could package yours up altogether so are we going to have an s203 driver board anymore now that we're picking off the shelf display this need to be moved to one of do I thought that yeah no there was a there was a different task that within the last sprint it was single but uh I wrote a display driver for a lcd or a led to phase display 15 20 years ago the guy sent me 15 of them and I asked him why and he said you'll see and I burned them all out it's not a trivial issue so yeah push it down the screen push it down the screen worries we're not doing it we're picked it off the shelf so yeah we well yeah it was uh it was actually quite an issue with the xilinx design that we didn't get the rights correctly on the bb interface and the scope you know scoping and checking voltages blah blah blah blah okay do we have an epic for milestone zero we do not kind of do other epics we have an epic for everything but is milestone zero and one I mean like milestone one is mocked to decay credit right yeah we we uh we started calling it milestone zero because uh we came up with the it's the intermediate milestone of getting a mark 2 to l the microsoft employees can we just drop it into the same epic go what do you want to split zero and one out the milestones in the epics are not interchangeable so uh I haven't figured out how to do real milestones in JIRA yet so issue for the zero versus one okay uh so do should we just look at what's in here and see anything else in the mark 2 decay prototype should be included in this sprint um I was just looking for any mark 2 tickets that are not in the sprint and there aren't I mean there are literally no mark 2 tickets that exist that are not already in progress it is good what about four speaker drivers is the back wall are you sure yeah you can put that in the sprint I mean it's already I'm already doing it it's I'm I'm active so this sprint yeah oh well that's a different problem you have access to other mark 2 issues that could just be me not good information right source 4.3 display with the included dsa that is that your search for a display right that's the the search for an affordable one yeah reasonable display okay and you're working on that now so we can just be in the current sprint yeah yeah we're not doing that reorganized documentation in the repo Michael was that fine finalized mark 2 package that seems like something it's gonna it'll be later how do we just find the mark 2 package uh I didn't say much else so I think you know we took the depth packages so I think that is pretty supposing just based on the so what can we add a description of this we don't know what's what's this supposed to be I think it's a depth package oh dead end yeah sorry you said bad but I heard dev for this will be so that that is you know like if we go around that distribution mechanism then the red dots are bugs I assume correct yeah okay so this is a later milestones we leave it leave in the backlog I put at the top of the backlog because it has to do with mark 2 yeah but finalized mark 2 distribution package I mean that shouldn't you can be in the backlog but it's technically not really a backlog it's you it's a future sprint it is but we don't have the next sprint defined we won't get to in the next three weeks I don't think oh I thought all right all right I thought we were moved what's app get install microth got a curiosity that's that's uh you know the creating a a way to install microth easily on any Linux systems that you zapped yeah we should mark this as don't do it sounds like a weakness yeah well it was the original it was the original uh request that ended up with us right our solution to this yeah why don't we just say one two for now yeah why does it say add audio test wave file to image I think I think we have some dot wave files in our standard distribution what would be the purpose of adding another one I'm not sure you knew that existed when we wrote this ticket well let's see I wrote it um I mean I would like something more specific for testing I mean we've got like some oh yeah we like that some of those songs we want to play the test of speakers out and all that kind of stuff too right yeah there was some concern about that like I'm interested when we release it publicly that we don't have you know rights to it so it's been a bit I need to find a song that we can that's you know no copyright problem but also mark one singing songs well I would like to find something that actually specific for like some audio related things that's you know different frequency right I'm gonna sign this to you then so you can do your research and I'll put this at the top of the backlog because it is mark two related but initially just to see if it works yeah I can just use like the alarm beat or something but once we get past right are we just printing the backlog right now I'm just looking for anything mark two down here is all I'm doing got it got it I'm not going to hear a song back in black I guess that means you should never interrupt ACDC that's right I'm not going to do anything with any Kiwi or GUI issues yet because we do stuff to make a decision there uh let's see test SJ 2 1 PCBA fit oh it's done yeah I just saw a couple things like that I've already did that I don't know I don't get I need to figure out a way to filter duns out of the backlog I've changed like that should belong in a future sprint yeah the audio performance stuff yeah we just don't have anything we're not far enough along the test that kind of thing yeah so is that top of backlog or is that the next sprint I would think next sprint not going to get through at this sprint no so I guess eventually we'll need an integration test for all the mark two stuff as well right let's remove pedacious fallback skills some of these things just just seems like they're creating a large list where we don't need one this is basically our dumping ground for everything and eventually we could do stuff like this and go through and try to move stuff up uh because this this backlog falls in tickets from every other project yeah yeah so it's everything that you can say right there yeah but my I guess my question is is it our intent to go through this backlog every two weeks no right now I'm just looking for mark two stuff that we might be missing from our immediate um sprints on the yeah you see something that's not related to mark two just don't even bring it up um but it looks like we'll probably need to do some vk stuff from for the mark two right um before we ship it this great feature files for skill on mark two didn't get finished the mark two I think this speaks to whether the vk tests are mark two specific or generic right now they're mark one specific we're building we're setting the platform on the vk images to mark one because it's going to be my closure what question is should they be enclosure specific would seem there would be a set that is generic across all enclosures and then the enclosures could have enclosure specific tests but yeah I don't know that it matters to move it out of the backlog or anything all right I'll put it on top of the backlog I'm going to have to think about that mono versus dual speakers in dev some point down the road not if we're going to show all the issues it's going to be a zillion you're going to have to filter by money related stuff and I've gone I've gone through the mark two related um tickets only and there's you know I don't know there's I guess there's only 45 in the backlog I guess but there's um but really none of them pertain to the next sprint there is an issue now let's see here yeah mk2 116 uh was something that we contemplated uh planning for the current sprint um I think it's basically work in progress um at this point I'm not seeing it there it is well yeah so I think just I can assign this to is this the same thing that Ken's kind of working on right now is exactly is it duplicate can I take it for this yeah you can market this duplicate okay yeah I was gonna say that's a duplicate right there well there's there's something to discuss with this I mean is this what we're calling enclosure code right for the new sj201 yeah essentially yeah okay well so we are blocked are we just quick question Michael so the conversation over the weekend with switching to um you know i2s from usb like with a new firmware update that will affect our closure code right yes it will so we're kind of blocked until we just decide how that's gonna work with what we have now if we can okay um I don't think that's more regardless because the amps we have to get the amps working which is about it's not in the system right now I don't okay I don't think it affects it I mean if we the things that Kevin's doing for the uh noise cancellation the uh the i2s buff doesn't affect the code in other words it's not a code issue so whether he fixes it or not code doesn't change well I thought I thought I had to do with how we're selecting a playback device possibly or so no it it it has to do with the fact that we were going to try to do it in software that's probably not gonna with not even in software in a configuration change it's probably gonna have to be done in hardware but I mean yeah I don't I don't think there's really any code change here if Kevin's not able to get it working in hardware we might be able to cut a ticket to say try to get it working by combining the two channels into one but I don't think it has to have a ticket outstanding that we revisit and ask what it is every couple of weeks or whenever we go through this well so if you're saying that's not really blocking us I mean we could technically we could technically have you know the enclosure code working and have mark two is working without margin and you would have to use the line out on the the board as opposed to internal amplifier no no no no this has nothing to do with barge into the best of my knowledge and it doesn't matter about line out or not what it has to do with regarding the i2s bus is pulling the noise from the speaker out of the microphone input and that's that's pretty much a hardware issue that he's dealing with and yeah it doesn't affect the software the enclosure code will not change one way or the other whether he's successful or not yeah but what i'm saying is we won't and that is a hardware requirement for barge so we won't be able to do any barge until that's resolved why do you believe that's tied the barge in uh because it's it's playing um you know it's playing the output back well my understanding is you're trying to match up the out the audio output with what we're hearing through the microphones and then canceling them out you know yeah but i don't think it has anything to do with barging is my point i i think bar is a different concept well that's we have we have a listen up while we're playing output um you know things like that barge in its own beast and yes but the cancellation of audio the aec is is a requirement for barge like a hardware requirement for barge but we can't do yeah we can't do it on the mark why we didn't mark one has no barging capabilities whatsoever it's too loud it's too loud it's hard to i don't believe that to be true is my point derrick i i believe that barge in is a function of refactoring our code in other words even if this were working in hardware barge would not be working today because our code doesn't support it we already talked about this i mean i don't support it in some cases like music playing right there's issues you don't have the ability to do it and then there's some software changes we need to make you can barge into a song playing or something yeah you would have to do some really really really tricky stuff to ever make a mark one work with barge just because it can't when it's playing something um it can't hear anything it can't hear anything besides itself the microphone physically can't um it's just you know the the music or its speaking itself won't allow to hear anything so you can't interrupt the but what i'm getting at and yes chris you are right there is barge in that's working when you're playing an mp3 when you are playing a dialogue barge in will not work with the current code base regardless of what you do with the hardware so right so we're talking maybe we should rename that with rename that to uh hardware uh hardware barge and software barge yeah it's with noise cancellation it's hardware noise cancellation well it's output cancellation you're specifically selecting the output from the input in hardware that's fine that's fine however you want that's technically correct however you want to put it but what i'm worried about is that we seem to feel like if this i2s problem gets resolved our code base will magically support barge in overnight and the reality is it will still not support barge in while a dialogue is being spoken until the code is altered to do so yeah no i okay well let's call it uh eight yeah you're right um i've been calling it barge in just because we can't do it without the hardware being capable of it which is the specified hardware barge in support yeah or a actually eight we don't save a hard dream here okay i mean you can go ahead yeah no we don't have to save it we'll call it one it is it is actually called an acoustic echo cancellation that's what is necessary acoustic echo cancellation is necessary hardware component to get good barge you need one one is a prerequisite today um for i mean okay i don't know like i you know things can be done software but like i said you'd have to be really uh it's not gonna be that's where i was going is that if kevin can't get the i2s noise cancellation working i bet you i could still get barge in working while a dialogue is being played yeah but the problem is getting from that would be so crappy it would be worthless yeah and you're not going to do it i don't know that i don't know that they're independent like you know because because the audio because the dialogue is not going to be getting spoken at too loud of a volume necessarily then potentially you could but it would also mean i mean yeah like so yes yes you could i think you could do dialogue interrupt without a c and by the way we also have a uh what's it called active or activate or whatever our action button that could be potentially leveraged to do barge in by pushing a switch that's how the mark one like if you scream at the mark if you yell at the mark one while it's doing stuff then you can like technically barge in so like you know over music sorry you can you can interrupt it over music if you yell at your device but it's it's not a good experience at all and like yeah all i'm saying except who is i i consider barging to be a separate issue we can we can talk about them in the same breath that's fine okay yeah let's not argue about the terminology if we want to get more specific about the terminology then we can make that an offline discussion yeah well and then there is the fact that you know for us to get what the xmoss promises then we do need the i2u to ask for i mean that's part that's a big reason why we're using this xmoss ship is because of the aec capability and it's exactly what everyone else is using not necessarily from xmoss but from some dsp provider to be able to do this kind of communication while it's playing music so you know we've got to get this thing to work that's why we chose okay moving on um investigate why we get stereo audio on sj201 is that's going to be done the next sprint or two or not uh yeah we need to do that okay so that's going to this sprint or next sprint um let's do it in this one that's something we should work with Kevin on okay sj201 test jig we don't need that won't do no not won't do but sorry it's not it's not this milestone it's a future milestone okay is this a that looks like a duplicate it does look like a duplicate it says sj202 oh you're right huh it looks like a duplicate to the untrained eye so the 202 actually you can um our requirement stock are we is there were we gonna write one of these oh look it's i have something written from kind of a design perspective or from a product perspective rather but um you know we don't have things like engineering requirements you know like you know we need to be able to far field it you know 10 yards or you know things like that um yeah so anyway i i think it would be a milestone zero was that just this feels like the the hardware requirements for milestone zero yeah well i think i mean some of this stuff's kind of nebulous though because we don't really know how well the sj201 is going to perform right so these are and in some ways yeah just like Derek said i mean this is more of a user experience kind of thing and the you know what are our expectations as to the performance of of the device right so what i have is the user experience side so i'm like you need to be able to you know interrupt it while it's playing music you need to you know be able to have good quality sound you need to you know i just elicit all the things you know we need the buttons for volume up and down and action all that stuff but in terms of you know the performance requirements of engineering aspects of yeah we don't have that right so what do i do with the ticket um let's put it into the sprint 17 yeah i think in some ways we have to like actually have a board working and then we kind of reverse engineer what we want it you know like is it working to our you know we can test and say does it work at 30 feet you know you know say oh it's not working at 30 feet we need it to be better than that okay can you're looking at this so you you want me to put this in the current sprint the enclosure and version core version yeah no no no that's that should be i don't know why it's in the backlog i'm working on it or you know i spoke to you about it and i'm yeah but i'll uh i'll get the enclosure code to be updated to put the new stuff in once we figure out what that last value encoding is but yeah yeah put it in the current sprint please okay uh system diagram for SJ201 based mark two i think it would be really useful like we keep referring to the flow but if the if the circuit's gonna change you know that'll no need to update that right now at the bottom there was also a debug the project rollover device uh sign that to me or derrick um and put that active i mean that's gotta be at the top of that might be a duplicate i thought we already had one of those i don't know but i wouldn't see it in backlog so these are gonna my radar or derrick's radar yeah i still haven't received anything yet um but yeah as soon as i get it hold on i want to do the right thing here yeah here we have this bug here right so that's sure is that the same thing no it's no these these are dead dead device like we're not sure exactly why well yeah i mean this speaks to the issue that we're having with the re-speaker board okay it may be related but one of them is it uh sometimes rebooting at a number of times will fix the issue and this one seems to be no matter how many times they reboot it it doesn't work yeah okay so these are different i just think we need the rollover one for now okay and ken is working i'll just put in the sprint you guys can give it to me or derrick whoever um it's kind of derrick for now um we should probably get this done too shouldn't mean the misspelled method name just or do we do we tell them what to do or i forget where we left this yeah so that just why don't you um it's it it's in the existing skill code or whatever the uh interface dot pi i mean i don't know how you want to get that out there too okay there's a comment here we're gonna i was gonna make a change in a branch that they can uh disable update we talked about it's an active high skill then you can update it in the master and and uh change the isn't isn't that also in the micro skills repo and it'll update automatically on their device no i think there's a there's a kitty branch to that though if they're probably all right all right i'm pretty sure you can delete mk2 138 it is check with michael but i don't think we have the ability to update the firmware on the sj201 it's marked as done so yeah the answer is yes delete it you know whatever uh this disabled update we were just talking about this earlier is this this is a communication issue right or is this it should still be in the spring it should still be in the spring it's assigned to me and i'm gonna assign it to derrick if it's not i mean if it is a duplicate then delete it but if it's not what i have to do is i have to update the ticket with the exact configuration change that has to be made and then i'm going to assign it to derrick communicate that to the account the account so now okay so i think that's everything in the backlog related to my tomato um milestone one or milestone zero is that so some of this is at the top of the backlog and some of its friends all right so the next question then is everybody uh packed full of work for the sprints uh my answer to that question is yes um yes and you got plenty plenty to do in here for two weeks all right ken are you uh you booked solid with the stuff for two weeks uh uh yeah i mean if i'm not trust me i can find something to do it doesn't seem to be a problem certainly the intense stuff would consume me so yeah oh no but that's just a review yeah i think i'm okay if not i'll raise my hand okay derrick you've got comfortable with what's in here for you yeah yeah all right i'm going to start the sprint then any objections i just want to discuss one thing briefly um after we stop recording are we really uh giving everyone that entire experience i wrote it down no it's just i don't know if it's sensitive or not i just error on the side of question yeah sprint created go forth and sprint thanks everybody