 All right, so kicking off the July 27th DevSync meeting and All right, so today we're kicking off a new exciting project in starting to add some more community features around precise and We're laying out the roadmap for that. So I think there might be some discussion about that today and We've got Well, I'll just start with doing my update On the hardware side, we should be getting back our first rev of the boards this week and So that's pretty exciting. In fact, I think we actually got the first of the boards today But there was an error on them this and they actually came so fast that they shipped them before he could fix the error on Saturday morning, so So he he sent out another order on Saturday and they should be here in a day or two. So that's pretty cool and So yeah, maybe by the end of the week we with you know, if we're really lucky we might have a working board, but that seems unlikely We're officially kicking off our Fundraising process today. So for all the investors out there if you didn't get an email from us then You know reach out and we'd be happy to talk to you And other than that, I think we should just get into the regular updates Let's try to keep it a little short. I was noticing in reviewing our notes from the other day that We tend to take about five or six minutes each and maybe we can Just try to be a little bit more brief and save some more time for the discussion at the end so That Let's kick it off with chris guesling Damn, I hate going first but um Yeah, so jim will work around the The service startup this week Which is transitioning from ugly towards good. Um, so Merging the services. Okay Did which tell you whether the services um Started or not will need to iterate to improve on the time, but um, that's working well And We've got some new tts festival tts support coming Is that time to do a release of microchore, but um, you know That in itself is a bit of work. So I haven't put time into it. Um yet And I should have been more prepared to jump in earlier. So That's me okay Next clockwise around my screen is chris fair Yeah, so the good was I went through our Bug fix sprints and cleaned it up a little bit make some things that were bugs But not marked as tasks actually bugs and move some stuff out of there There wasn't really part of that sprint. So I moved some things around And while I was in jira, I created The epic and all the tickets Um For the upcoming precise projects Um, so we can go over that and now we want to organize that later, but um, that's all done Went into confluence also and started writing some documentation On um, what the tools we're talking about building will do and what they need to do And how they will do them. Um, which I hope to publish in the next day or so So that was all good. Uh, nothing Really bad, but the ugly being that I'm still in um, Maryland So, um, I'm not getting as much My velocity is not quite what it usually is Okay, understandable Um, all right, thanks Derek all right, uh, so mostly just been working on project roll over stuff Um, not a whole up on the mark two hardware. Um, other than what michael talked about earlier Uh, I can show you guys What I will be sending out tomorrow Um, so here's a few of them You guys see that? Clear one very nice They are They're kind of semi translucent. I was going for that late 90s kind of effect That's actually, uh, the material is is a little bit more durable than um Some of the other stuff so if it like it's dropped or something they will hopefully last The other stuff I use in the past was a bit brittle so, um Yeah, those are Pretty much wrapping up and charlie's I told him to just keep working on it. He's instead of attending the meeting. He's Wrapping things up as we speak So, yeah, that's all good. We did the testing heat heats fine. We tested over the weekend Didn't have fail. Uh, I ran it through just the simple skills tests This morning and no problems So i'm pretty happy with that um, so he's got to get these boxed up and I need to do a little bit of testing tonight and then box them up So they should be going out the door tomorrow um Yeah, and then hopefully once we get those out the door get back on um A uh a quick enclosure for when we get the first sj prototype it'll be laser cut for the first one So that's it for me Awesome, all right Let's see, um Oh, no the order is completely changed. Okay, ken So, uh, let's see. So I I don't know about good, bad, and ugly everything's good. It's all good. So, uh, I created a, um Kind of a roadmap wiki confidence page for the precise project as I came to understand it Cross linked it with the what is it precise 59? I guess epic So perhaps chris for your documentation to make it easier. You could just inject it into that page That page is basically ordered in order of execution or belief A perceived order of execution So if you disagree with that that would be a good place to surface that and if the products you think you're working on are different than the ones that are Anunciated on that page that would also be a good place to kind of call that out I think we're in sync though because um, you know, I did that Without realizing you had done the epic and then when I tied it back to the epic I can see where the tickets would fall into that page So I think we're in sync. Um And I don't want to do a summary there But I'd rather put the details like all the design and stuff and a different page for each one and we link those Because that will become massive. We put all the different things in there So sure and we can keep this more as a high level roadmap summary. So that's fine And yeah, and I'm interested when you do your detail, we can continue to have our repartee regarding trade-offs and design issues and things like that I think you've noticed already we have a A little bit of a disagreement, which I think is healthy regarding how to go into these projects I'm always looking for the how do we leverage what we have existing and prop it up and then Cut new tickets for enhancements And you want to take the opportunity to kind of Make them better and look at all of the things we know about upcoming And try to potentially get them into a rearchitecture And I think that that's a healthy trade-off there that there's good trade-offs in either case So we should probably keep that line of communication going And again, as I mentioned, I updated some of those tickets as well with some of my input I isolated the data To be deleted and I'm ready for the go ahead to push the atomic button, which I got earlier in the meeting So I will whack that data ASAP. I provided everybody with a more detailed Uh analysis of what files and what accounts are being deleted It probably is a csv that contains about 75,000 rows as opposed to the one I sent out last week Which was the summary rollout by account, which was about 207 rows So between those two csv files, you should have all the information necessary to reach out and communicate to the community About what's happened and what we're planning on doing So I just you know, I also have a document in the works Like chris, which I haven't published yet Which is a kind of a presentation overview Of the entire precise process and ecosystem if you will Which uh, my intent is to merge that With the document that josh sent me Which was how that used to be So basically to take what I've got integrate what josh sent me and combine them to become current So that's the last piece of documentation. I'm currently working on was able also able to get a little work Squeezed in between all of this documentation so I ran an initial set on saturday Of cheese, I want to say 15 or 20 models and resultant performance output Against the high and low pitched test data sets Where I started looking at the hyper parameter tuning but from a manual perspective And there was two reasons for this. The first was to validate some assumptions That I have which is a we need more data and And then the other assumption is that being more epics would serve us well and so There is some lip service to that in there. There's two training data sets that are vastly different in size that are used in that Output data that I posted on that ticket or not ticket but on the hyper parameter tuning wiki page and then I also ran with the concept originally from what I've learned regarding the relationships between batch and epic size and between dropout rate and Recurrent units And you can kind of see the logic in there. I mean because it was manually done So you can see where I kind of hit one I try to hold like all hyper parameters constant and just vary one And go up and go down a couple of notches and see if there's any improvement And then try to run with that So I think that that is somewhat self-explanatory But it's it's just scratching the surface and it's a manual analysis But I wanted to get some variants in there and see What hyper parameters had what effect On model performance. And so I've updated that wiki page with that It can be found from the main precise wiki page Under hyper parameter tuning And that is it for me. I will complete or try to complete this documentation this week And get like I said the data deleted and then work with chris on Some of the architectural trade-offs and things that go into propping up the Uploader and tagger into the Continuous integration environment we have called Selene. That's it for me Okay, thanks Ken Josh is here Just figure I give you a chance to say anything if you want No, it sounds like people are making good progress. I'm glad to see the the Privacy issue being resolved. I'm writing an update about that um the I've got the ball with the um The customer I'll reach out and and have a chat with With rollover and make sure that um that they're They understand where we are. I do have some input which I've dropped into the Into the chat about how to organize those drives voyage Linux does a really good job with it. Um You know, I've I've voyage machines that have been on for Actually powered up for nearly a decade with with no problems at all that are using compact flashes their storage mechanism It's a the trick is to treat flash differently from the way that you treat a regular hard drive in terms of How you're organizing the file system and how you're doing read and writes um I don't really have anything else for dev sync the european union's about to carry some water for us. So that's really exciting and lots of other Good things going on really happy to be Pushing forward with with fundraising again All right, thanks um, so, uh, I noticed at least one topic come up that we should really Discuss and that was that uh, we haven't done an update in a while and um currently those updates really just go to the mark one and I was wondering if There's anything that's preventing us from doing that um, especially the migration from Uh, the to the major versions of core that we've been doing I know we we did an upgrade when we went from the the pi 3 base to the pi 4 based, you know Model for the for the mark 2 and I was wondering if that has any repercussions with respect to the mark one I don't know about the mark one mark one, but my input question is When we upgrade Minecraft core Does that automatically get pushed out to every device or is it a opt-in update? Minor updates Are pushed out to everybody major updates we ask The only reason I ask is Yeah, the only reason I ask is nothing is more annoying to have a working system Power it down get up the next morning and power it up and it's no longer working and you didn't do anything So that would be my only concern there Yeah, so the the minor update so, you know any 2020.02.something that that's the minor update so that should happen automatically And then any switching from 2002 to 2008 That will Comp the user and say hey, there's an update available. Would you like to update? They say yes, then it goes through that update process We do Whenever we do a release of microp core, we do a lot of testing On iCroft mark one And linux and you know now mark two But we yeah, we do that across those devices to to make sure that it's gonna Well to hope to do a final check so obviously the ci process should be catching anything that That is a problem, but As part of the release process has just been doing that manual check as a as an extra Stacey mechanism because you know exactly as you say Ken You don't want to have your device stop working when you didn't do anything a bit to it So Yeah, I think uh It takes You know the better part of a base to one of those releases what I think it'd be worth doing a doing a release at all Just to make sure we're clear when we do minor point releases There's no deprecation in that release, correct? Correct, yeah Only the majors just All the minors have to be backwards compatible And so any skill that works on You know 20 or two any skill that works on 20 or two should work on every minor minor releases 20 or two um, and so then when we switch over to a new major release and the skills repo also switches over to the new major release Of the skills repo and we test all the skills to see if they Work on the new major release and if they do then they can be automatically um, you know put into that Into that new Branch, um, and if they don't and they get bumped and we we let the authors know that you know There was a problem and and invite them to fix it up so that it can get re-added And we do release notes for major and minor releases Yeah, yeah Okay, so our our Our typical schedule would have would have been last year for example would have been to put out a 2008 Release is that right? Yeah, okay, and so that's coming up pretty soon Um, do you should we schedule that now? Um, Did you do it on the first of the month the last of the month just some convenient time in the middle? Apparently by the end of the month, okay Okay Then let's get that on the calendar so that we can account for it. That's going to take you a day or more And yeah And so that process like I like I like I asked uh, so that does not affect our ability to um Because another were some issues with the the wi-fi setup and things like that that are different between the mark one to the mark too right They're independent, of course, so they shouldn't shouldn't be affected. Gotcha. All right Okay, but I would like to get this availability stuff into into Um 2008 for sure. Okay. I don't know it shouldn't really be a breaking change But just in case there is anything in there that we do need to right. Yeah, it's a fairly significant change though so Okay, great. Um, was there any other discussion necessary on our precise roadmap? Uh, sounds like you guys had a good discussion before I was able to show up Um, mostly that the to me and just buy off on in your IU that that's the scope we talked about Um, and I didn't do any kind of scope creep by Putting any of this anything in there. Okay. I I tend to do that because I want to fix it all right Uh, okay. So you guys have all discussed that though. You just want you want to get my input on that Is that yeah, I mean Ken and I have been back in court a couple times today. You know scope So I guess in the bottom The two two-minute version is that you know Do we Just prop up what we have as we have it and then fix it later Or do we you know make this Like the schema for example, do we take the do we take the scheme as it is and Or do we really um, you know Look at it. How does it really fit well in the salini? And you know, so is it you know, it's like a quick and dirty or is it a let's put a little extra effort into it and Make it a product, you know instead of you know, there's something we can maybe sell outside of you know core Gotcha, okay. I'll take a take a look at that ASAP, uh, so for this week everyone is still um making progress on the bug fix sprint Or at least for today Tomorrow until we resolve moving off this bug fix sprint Yeah Chris if you can have a look at the oh I wanted to get your input on that um that service availability stuff because you said you had thoughts on that and I I did have a sample pr Um with my approach, but you know, I'm I'm sure it will differ Significantly, so I just thought a sample pr was an easy way to discuss it, you know Okay Yeah, I can take a look at that Okay, just taking a quick gander through the uh the sprint here. There are nine tickets left that have not been Uh that haven't been started anyway in terms of progress only one was marked as a low priority that we can just boot off the sprint um So and then we've got three in progress So at the rate that we've been closing these tickets and just taking a Uh You know your gut feel for how long these tickets are going to take to resolve Um, what do you think does anyone have an opinion? Uh about how much longer we'll be working on on these Oh before we do that I mean I went through those tickets and I think we said to move Everything that was important to a like a very high priority and I I don't think there's anything in there that's just earth shattering right now. I guess is my point I don't think anything is in there. That's just gonna it's going to keep project roll over from going forward or anything like that um So to me if we want to start, you know moving forward on the next project and You do another bug fix sprint after we get the precise stuff for about I would be fine with that. So that's that's just my opinion. I don't know if Chris or Ken if you've been through that Sprint they're not I agree with you. I I think if you can get started on uh, you know, um propping up the or migrating beginning to migrate the existing upload Wakeward process, right and the scheme associated schema Uh, you could probably start getting working on that right away Uh, because that's really kind of the number one thing we want to get going, right? And uh, you know, if the rest of these bugs are low priority, they could certainly wait for a future bug sprint So I I would concur with that. Yeah Okay, yeah, I just realized as you said that Chris that um, I had misinterpreted these little orange arrows as red arrows So I see they are at all now medium priority not high priority. So So we've got three in progress by Gez And I imagine you should close those out Uh, unless they're giving you trouble Gez and you want to Uh, put them back in the queue They're all the same thing. Um, which is what I need to get Chris is input on. Yeah Okay, yeah Okay, great. Well then, um So we've got Do we have a new sprint for the For the precise and salini roadmap I did not create a sprint because I didn't know I kind of wanted you to Um To see it or see what was there and see if you had any input on what do we do first? I think We could talk about that right now. I can create a sprint. I think the the most basic thing we can do is get the um precise data You know get get the database work done and get the api work done Um, the front end stuff can happen as a secondary thing Um, but I think just just doing that will be a good first step Would you can you think that was a good approach as well as well? Yeah, I can even uh expand on that just a little bit. So yeah, I think the um The very first thing like you mentioned is migrating the upload server Getting it running and salini and into the continuous integration and deployment process Um getting the schema over there. Um, you know, you and I are looking at uh, you know Does it make sense to just prop it up as is or do you want we want to you know, uh, enhance it a little bit? That would be great. And then number two, I think right after that coming right after that Would be the resurrection of the upgrade precise tagger That was developed that josh was discussing and I think The issue is that it's not throwaway code and we're going to do it for the third time I I tend to look at it more like we're going to pay down some technical debt and get it integrated into salini um because it seems like the Salini project had to do some triaging to Get moving and get the tests going and all of that And now might not be a bad time to rally back around and say look at triage And let's see if we can pay that technical debt down and get that functionality back So that's kind of the way I'd like to look at it I'm very much on board with finishing finishing precise putting it into full production as a actual software package and Putting it into maintenance and we've been goofing off with it for several years now and We are not Inventing anything super fancy or new at this point. Let's let's finish it out. I move on to the next task We have other things to do Okay, so it sounds like There's a couple of outstanding questions. Maybe uh in terms of the scope So I'll take a look at that and give you some feedback Um, but in the meantime Um, you all have useful work to do Uh, that's just my main concern, right? um and Okay, so let's go ahead and do that. I mean we're we're moving forward with this precise, um You know productization effectively, right? We want to get the community back involved and and contributing to our our data set um and Oh, yeah, so let's get going on that. I will um, I'll take a look at that uh today and get you some feedback so that that's not holding anything up All right, awesome. And then I'll uh, I'll go ahead and create the first sprint with the api and database stuff in it and then if we, you know After that we can decide, you know scope wise what we want to do to make it to make it finish to get it completed Okay Do you guys need a description of what the The boot process should look like on this is that but do you guys have a clear picture of what a stable compact flash Operating system looks like or would it be helpful for me to just write something out? I'm not sure exactly what you mean So for I think to run be helpful to get a write up of what you're thinking So again, so to say how how the current process compares to that Josh are you Looking something up No, I said sure thing. Oh, and then I unmuted myself. Sorry Order of operations issue there, uh, okay, uh, and you don't even have the extra Oh, sorry, my one other thing is um, just if we uh, with the decision making around the the duet stuff, um Do we want to keep pushing forward on on that so that we can kind of do some testing and and Keep gathering data and how we're going to make that decision. Well, actually, I feel like we've got a To start a pretty decent start on the um the criteria to be more populating the you know classic pros and cons table of uh for each criteria on on that Okay, is Is making a What is making a choice on that blocking us from working on? Right. Okay. I'm just wondering how far we can kick that hand down the road Well, I mean Would we spend time on both sides of the fence like every single week? so It doesn't block anything it just I see Hmm And this is because people are developing skills in for the mark one Well mark one doesn't have a duet so Yeah, so the community community developing skills for qt We're doing this project rollover in kv I catch up Derek and I catch up with blue systems weekly um and Try to unbox them on qt things But you know at the moment i'm intentionally not putting much effort into that qt data things because Like, you know, I don't know which way we're heading, um and Yeah, so that's Difficult there because you know If we were actually pushing forward on it, then um, you know, there'd be It would change priorities I think it's I think it's past time for michael for us to connect michael with the leadership over at blue systems to Have a discussion about that like they're putting real resources into it and we really appreciate You know all the work that they're doing At one point there was talks of having a much closer relationship with that organization and you know Through a variety of missteps on our part and possibly Some communication issues related to languages You know those conversations ended You know given that we're You know going through a fundraising process right now And that we're making big decisions about architecture now might be a real good time to connect with the blue systems team Yeah um We I mean as you know, we're already Intend to connect with them on On other issues. So yeah, absolutely. We can discuss that as well um Okay, and similarly for the wi-fi setup, right that's not blocking us from doing anything Aside from Yeah Okay, right Um, all right then so any last Words No Okay, great then the I guess we've all got our plan for the next couple of days and I'll see you in on thursday Not before Hi guys