Collaborate on Lunch Bill Format for Building Services with Ease Using airSlate SignNow

Lunch bill format for Building services

okay and we'll actually paste or uh we'll keep the music going for a second let me get a bunch of these links together and really I need just like a page of this stuff no I didn't work at stripe and Google why would you think I did that oh you asked co-pilot yeah no yeah co-pilot's wrong about so much oh where does it say that like it was like an auto completion or something where was it showing that oh yeah I definitely don't no no yeah it's totally wrong about all this stuff nah definitely not yeah it's super Maven co-pilot would do the same thing though oh that's the wrong one um I'd say connect. te I'm not going to put that one on the list just because I think it relates a little like it's basically the same hold on not that one this same topic so I don't know eh I still think this is like hugely relevant but yeah you know uh so yeah squiggle comp I'm a little annoyed that the squiggle comp videos aren't public I I think that's a little odd cuz it has been two months you know but yeah I think one of the organizers is probably like editing it himself so I could see it being some effort but yeah I was really hoping that the videos would go live for okay so then bimon so yeah that's that one oh I forgot about the uh custom short codes one yeah let's add that one in I feel like this one was live I don't remember pre-recording that so that was Hugo comp 20 22 um and real quick um what's up Adam how you doing out there thank you for the follow all right so neovim comp three years ago interesting so they didn't publish mine H maybe because it was Fuzzy yeah I don't know that's a little Annoying well I'll just remove it it's easier that way and then Manning rust now this was over in like the IMAX studio so little different okay so there's all those talks and yeah I'm going to save this somewhere I don't know where I'm going to save it though just put it up here on t7 and call it that all right uh not yet no what I need to do is I need to record some videos for these and then we do more sass yeah let me get up and use the restroom while this song goes because it's such a good song and then I'm going to turn the music off while I do the recordings for talks e okay and actually let me double check I might be able to record and the songs don't make it through I can't remember I'm pretty sure actually but I want to double check that so yeah starting this recording just seeing if uh if this works right me you know making sure that the uh audio from the music I'm playing doesn't make it through all right show the recordings folder hold on we're going to let it autoplay let me just mute Chrome yeah we're good there we go okay uh no it's not making it through to the recording that's what I'm saying B Nintendo I have it set in a way that it doesn't show up on the recording I just wanted to double check that before uh before we did any of it okay so yeah we got just three recordings to make and then um then we get them uploaded and then we work on our actual project so let me just do a a rough draft real quick and microphone check microphone check yeah volume looks okay so hi I'm Chris griffing and I want to talk about how we migrated 600,000 lines of code probably more I think around 660k uh from flow to typescript at get Kraken our get Kraken desktop app is an electron react app and uh you know it's been using flow since about 2017 the project was started in around 2014 and we needed some type of type system to make our code you know easier to reason about uh you know same reason why today you would probably pick typescript in this talk I'd love to go into you know why we picked flow at the time uh a bit of the history around get Kraken desktop and you know when it was started and some of that and then you know how it's grown into this massive thing I then want to talk about flow and typescript and how they both kind of evolved separately around the same basic need and maybe some other type systems like reason that might have been uh relevant at least for a Time and uh from there I'd like to maybe show some of the differences between flow and typescript right and maybe why typescript one out I don't think it's purely based on syntax alone I think there was some not necessarily politics but just simply about how the approaches to these things were taken and you know why maybe typescript one uh I'll also talk about some of the uh challenges that we faced when migrating flow to typescript right it's a large code base uh Tech de was not prioritized until somewhat recently and you know like for the developer experience there was extremely slow type checking we're talking like 30 seconds because flow had to parse the entire code base every time you made a change to actually determine if you made a mistake that's a really bad experience where you do something save the file and then have to wait to see if you forgot to add a question mark to a like safe navigation operator kind of chain uh and then from there what were some challenges we had and how did we solve them in migrating that code base so I hope that sounds interesting to you and I'm looking forward to maybe speaking about it boom right I think that was around two minutes right it doesn't have to be perfect right how do we feel about that for recording I didn't record it yet I'm going to switch to the uh the other view here in a second to where it's just just me you know so yeah what view is that I think I just have to switch it manually yeah I don't like the fact that I'm getting cut off here like it's really close and it's cuz the beanie is taller yeah I might need to like move the camera up at some point we won't do it today but you know whatever all right so let's get this recording going oh and I see what I messed up there in doing that Ah that's just a little Annoying hold on let me just move it up just a little okay that should be better it's a little low now but whatever I think it's better than getting cut off and really I'm wondering if I could just raise the chair a little nope chair's maxed out I'm sure it's fine I could actually lower the desk a little bit I'll fiddle with it later hold on let me look and see how it looks here on like the main main scene I'd say that's fine I want to move it up just a yeah yeah yeah I'll move the desk down just slightly in eventually you know we don't have to do it right now all right so let me start this recording I think we'll probably just you know do a nice easy rough draft and it should be fine hey there my name's Chris and I'd like to talk about migrating our 600,000 line code base from Flo typescript at get Kraken the code base is for get cracken desktop which is a react electron app and it was started back in 2014 and we adopted flow around 2017 um I'm going to go over some of the history around why flow and typescript were around and created and kind of adopted by organizations why we wanted to use it some of the key differences between those two and maybe a little bit about why I think typescript one out I'll be referencing all sorts of things like the documentary about typescript and some of that information that's come to light about just how they took their path in creating some of this stuff uh I'd also like to maybe talk about some of the challenges we faced with migrating from Flo to typescript so it was a large code base right it was around 660 total thousand lines of code Tech debt was not massively prioritized for a while just because we need to ship features we need to be profitable and the type checking was extremely slow so we were dealing with around like 30 seconds because flow has to check the entire code base when you make any change so that can be really annoying if you make a change to your code and then 30 seconds later you get told you forgot a question mark dot for some safe navigation operator stuff so you know we were stuck on an old version of flow it could have gotten faster but upgrading to a newer version of flow would have been almost as much effort as just simply adopting typescript instead so I'll go over some of the challenges that we ran into with those things and how we solve some of them and you know basically some of the stuff around that and I hope that sounds like an interesting talk to you how long was that recording how do we feel about that I'm looking for any feedback you got feel free that was around a minute 44 perfect any feedback come on guys help me out here all right should I do another take what do you think no worries just just trying to find any feedback that might be relevant all right so time for number two who does number two work for hi my name's Chris griffing and I want to talk about the current state of State Management for react most almost every application of significant size is going to have to manage state in some way but how we decide to do that can be vastly different it could just be something as simple as context and using the apis that react provides you could be using Redux which is one of the oldest uh versions of State Management still going today and there's even some newer things which aren't as new and I'll dig into some of that like signals and um you know maybe Atomic State Management as well and what are differences in how these things work right the main gist is State Management allows our components to render based upon some State changing from outside of them but the mechanisms by which they do that can be very different so I'd like to cover some of the examples of libraries and tools that do those different types of State Management and I'd like to cover a bit of the history around why we need this type of State Management and the history of flux and why Redux kind of won when it came to some of the older flux implementations Etc uh and and you know I'll we'll look at some code examples and compare and contrast the way that you consume these different libraries as well as just you know list out various libraries you might decide to use for picking a certain type of State management such as signals atomics uh or context or Redux so hopefully that sounds interesting and I look forward to maybe speaking about it okay I'll call that one good unless you guys think I should do another okay yeah and I think that's what they're looking for for these videos I think they're mainly also just looking to make sure that people are able to like actually talk about these things you know like it can be really really off-putting when someone is not necessarily good at speaking right you know like I I don't know if they would really make this criteria but like a really really thick accent can make for a less you know uh a less entertaining talk I don't know if that's some of their criteria I don't think so but I do know that you know simply like when I'm listening to someone with a really thick accent it's a lot harder to actually like get the content that they're trying to put forth uh Jay so welcome we're going to get to the Sass here in just a second um I need to record like a two-minute like Spiel for submitting to react Miami first though but the basic idea and I'll give you a quick gist is that pretty much every other software as a service out there has the idea of quotas and balances so versel has a quota for how much bandwidth you can use or how many requests you can make to their back end how do you manage like or how do you record a user's usage of those uh balances or quotas that's something that I want to build out as a software as a service that smaller companies can have their own balance or quota management without having to build out that system themselves and part of that system might be notifying users when they're getting close to their limit or things like that right so the mechanisms for tracking usage basically right in the same way that you don't want to waste too much time building out your off layer so you might consider something like clerk is the same reason why you might consider using something that manages balances and quotas for you as well because anytime you're wasting on those kind of systems you're not actually building out your actual service right so yeah it's these things are basically infrastructure for your SAS they're not your sass itself all right so we're going to do this recording and that actually felt pretty good how do you feel about that for an elevator pitch was that a solid elevator pitch am I am I ready for y combinator what's up ker how you doing did you hear that elevator pitch just now krar if not no worries you've heard it before I think this one was a good version of it though four minutes of ads no way dude I've got that [ __ ] cranked down so much you shouldn't get more than 30 seconds of ads no no no no no no no no no no no no no no like what I've got that [ __ ] cranked way down how many minutes of video ads you want to show per hour 30 seconds done pre-roll ads for incoming viewers partially disabled uh add length and frequency oh they've got that on auto manual that was it so yeah they have another thing that's like a separate thing so ad minutes per hour add length and frequency shorter ads more often okay so yeah I'm not running any pre-roll ads myself but but if I don't roll them then that can be a problem yeah all right oh hold on so there's the ads manager automatically schedule video ads so pre-roll ads are now oh oh so yeah pre-roll I guess is just gnarly yeah sorry man I don't think I can change that I think ads over on YouTube might be better like uh Jay like how bad are the ads over on YouTube I'm curious about that though so yeah and what up padri yeah yeah it's a good chill end of the year um yeah I just need to do one more recording here and then we'll get to our code and then you know I could pour some whiskey and Coke ah you got YouTube premium Jay yeah I do too so yep I don't see that stuff either oh yeah dude um yeah I made the mistake of watching uh the streamer Awards over on their website instead of here on Twitch and like I guess I wasn't logged in through their like widget that was on their website so I was seeing like four minutes of ads every like half hour it was gnarly but more power to him get that money e uh and real quick hold on uh ads don't mind as long as it's not in the middle of something important get to watch this stuff for free and they do need to make money yeah here's the thing Amazon makes money off of like like twitch paying for AWS at basically market rate so Amazon gets to basically write that off on their expenses for the year I would bet that um a large majority of twitch is actually just writing off profit from AWS because AWS is massively profitable yeah like all the services that twitch is using it's basically a tax write off as an expense but they get to charge that expense at retail rate because it's potential like you know uh lost cost you know any any usage that twitch is consuming is potentially usage that a consume like another customer could be consuming so they get to build that at full retail at least for tax write off services or purposes for all right so yeah let's get this last thing recorded oh is there a calculator to see how much IBS costs oh and what's this what I get tagged on oh Jay was that you hey thanks for the follow so yeah let's just say we got one channel at full HD uh with how much I stream so let's like this is going to be per month right let's say I stream 24 hours a month which I stream way more I stream like 24 hours a week we'll say at an average of 50 viewers um but because that average is averaged out right that'll be 50% or 100% right what if we did multi TR oh multitrack actually makes it cheaper oh interesting and we'll say most of them North America but if we change region that doesn't change much oh other regions would massively change that o so right like dude it would cost me $220 a month to do my own hosted streaming or not not a month sorry a week because I stream at least five hours a day five days a week dude yeah untenable and I don't know what Advanced HD would mean but oh okay so if I only wanted to wait what just changed oh so at Advanced HD you're only at 720p which would decrease the cost a bit but at standard full HD [ __ ] that yeah so yep dude IVs is not cheap so just imagine right like how many streamers are going at any given moment right like 15,000 or some [ __ ] right what if we wanted to do real time streaming oh stream chat oh God uh real time streaming one viewers uh I we just say that 24 hours oh interesting market share of gcp and Azure I've heard someone say Azure is like a big market share but I'm I don't know Azure market share versus AWS this is saying Azure has 20% of the cloud Market where's the graph I got no graph uh yeah so gcp is around 11% Azure is at 20% and AWS is at 33% and then the rest is kind of like spread out elsewhere so yeah looking at this Amazon sits at a pretty solid 30% Azure okay yeah uh no yeah up to 20% lately gcp is just clawing for that 10% and then other providers like uh Alibaba is way less interesting I IBM used to be something and then others yeah like Alibaba and IBM are probably lumped into others at this point so yeah that's crazy yeah I didn't realize Azure was as big as it is that's kind of crazy to me but okay let's record this last little bit and this is going to be about forking boring avatars all right let me do a rough draft hey I'd love to talk about forking boring avatars boring avatars is a library for generating deterministic pseudo random avatars for users in a way to where that is done on the client side but it's consistent across devices which means that you're not shipping an individual like randomly generated image per user as an image asset it's just a 30k is uh library that you would add to your application and you know they're just prettier like more fun images it's similar to a gravitar but there are a couple different like Styles and flavors there's another tool called dice bear that we won't talk about but basically boring avatars is a reacton library uh there's also you know some forks for other Frameworks such as react native there's a separate library because of how the rendering is different there there is angular Forks there are just a bunch of them but they're all maintained by individual people and maybe like not maintained that much at all so what if you wanted to make cross framework components right you definitely want to consume it in your react apps but what if you have other apps at your company that you know might be written in angular or uh you know uh view or a whole buch of stuff so basically what we would end up wanting to do is writing a component that you could compile elsewhere and web components could be part of that right and why would you do that we'll dig into some of that in the talk but web components are not really that well uh accepted these days so nowadays there's another tool called mitosis which is pretty cool and you end up writing what looks a lot like some react code you get jsx you get context you get a whole bunch of The Primitives that you'd be expecting but it goes through and transforms that code via an a into a Json structure that it can then use to build out other framework stuff so you can build other framework components from this like intermediate Json structure using mitosis so I'll go through basically what mitosis is why not web components and a little bit about you know what boring avatars are why I decided to Fork it and a little bit more about like how that source code worked for the for including the playground SL demo pages that I needed to build out some uh mitosis components for as well I hope that sounds interesting because I think uh you know cross framework components are really important for people making Design Systems at large organizations as just one example right so yeah I think that was an all right one um and yeah I'm not sure uh surle they might have just been talking about I don't know what they were talking about maybe they're just trolling who knows but yeah yeah good call just deleting the message we'll give him another chance yeah exactly like you know context could have mattered yeah the little guy smiles and yeah if you ever end up wanting to Port it to another framework you know just in case you can definitely get it in view spelt solid angular quick react as well uh SAS is a software as a service there stoic so yeah just in case ker if you want to maintain that same look across your organization or whatever there you go uh software as a service is typically an entire platform versel is a platform uh but like stripe is a software as a service technically versell has plenty of software that I would say they still count as a software as of service right they're like orchestration and how they deploy things I would consider software Service as well I don't I don't care there stoic uh I'm going to start ignoring you if those are going to be the kind of questions you ask they're not sports teams man stop treating them that way all right so let's get a recording done talking about boring avatars we want to keep it around 2 minutes hi my name is Chris griffing and I'd like to talk about forking boring avatars boring avatars is a library for react in which you generate deterministic pseudo random uh avatars for your users based upon passing a string into it and that string gets hashed into drawing these things in a pseudo a random way the reason you would want to do that is maybe as a default image for users but having something that is specific to that user rather than just the same generic user icon for everybody and it ends up being like fully like cross device consistent as well because of how the pseudo random uh algorithm works but what if you wanted to use boring avatars in another uh framework or across your organization where you might have some other apps built with other Frameworks to do so you could maybe build out a web comp component but we'll dig into why web components are not necessarily the preferred way of doing things these days and we'll look at another technique for cross framework components using mitosis mitosis is a way of building components using a familiar react like syntax with jsx with context with hooks but what it does is it breaks it down into a Json structure using the abstract syntax tree that Json structure is then able to be crawled and turned into things like an angular component and basically like 21 different targets for various Frameworks uh and why would you want to do that so I'll go into some of the reasons why cross framework components are useful besides just the you know um large organization having many different apps approach maybe a design system things like that and I'll also dig into how I built my own Fork of boring avatars using mitosis and outputting to currently seven different Frameworks we'll look at some of the source code and even like the extra style that I added to boring avatars I think it's it's an interesting technique and uh I would really like to just show people the power of mitosis so yeah maybe not the best version but again I don't think they're going to get too upset about uh like a couple stutters here and there or anything like that so I think we got those recordings done let me tag them and then we'll do the upload process and submit to react Miami any feedback I could redo it if you guys think it wasn't good enough I don't mind oh and yeah honestly like this is the moment do you guys think I could have done that one better cuz I will or do we think it was good enough um based on your silence I'm going to say that was probably good enough okay so now I can uh get these Google Drive let me just go sign in think it was good enough cool thank you Okay cool so I can switch the right screen and let's do the first one all right H we'll just leave the CM griffin.com all that good uh get Kraken developer advocate okay and then short bio about [Music] me I not local I am based in the US have I ever spoken yes and then oh boy eh good enough um okay for e okay uh submit another response hit Kraken developer Advocate no yes yes all right so then this one is the current state of State Management for all right and then last one before we get to our actual code that we're trying to what's up standard pointer local oh I am in the US uh have I ever spoken yes and that and then the talk title for this one is forking boring avatars hey all right so we got our three three videos published we are good to go all righty all right so now we'll actually be getting to what we're trying to build and work on today uh we'll close some of this that was part of a different discovery and you know what it's Friday and real quick I missed some follows so ishmail uh gdpm and ref thank you all for the follows welcome um yeah might as well go use the restroom one more time one more again um I'm going to grab a bottle and and some soda and uh yeah we're going to enjoy our Friday so be right back one sec e it's Saturday where you're at well I mean Saturday is also a good party day oh that was a good Quirk pop all right we'll let this last YouTube song go it's just such a good Chill song quiet at the moment uh what about Cuba I live in Washington State oh a kual Libre that's just whiskey and Coca-Cola right uh Seattle Jaz was good yeah nice uh fun holiday event yeah Washington State mentioned and yeah standard why didn't you come out to Seattle JS no I get it it can be a pain like dude traffic was a nightmare on Wednesday no it's fine I get it and traffic sucked like it was really bad okay so for everyone joining the stream right now we can close that I don't need that at the moment basically we updated the CLI and the data r ER so I mean standard uh the second Wednesday of every month is Seattle JS so it's a really good uh really good Community I highly recommend checking it out yeah second Wednesday of every month there used to be a Belleview JS as well um that one hasn't gone for years though right pre-co yeah no worries all right let me double check okay so number one yeah that's right we got to bring all of this out into a Json file all right um and that'll just be over here in shared uh so this is data keys and we're going to have a lot of stuff to fix here and it doesn't really matter what we name it so we could just name it marketing emails I think that's fine uh the chilies so I mean in this file they're not chilies and technically it's broken now because it's not valid Json but the Chili's just annotate the end of a block so over in this code it's annotating the end of the block but here it can't discern what block that is just due to a bug but here we can see that this end of the block is that function user response whatever right or you know maybe down here we can see we're at the end of the block of if new user and user so if we wanted to add another block after it we get a little bit of context it makes a little more sense inside of like HTML files right like here is a better example right we're annotating this div element with the classes that denote it right so in like maybe a better like a better TSX file right we can see that the end of this div is that W full Flex this one is that Flex row so yeah it's Biscuits code biscuits and then yeah this part we just we could do just a find and replace too but it's fine all right so we get quader key there and then all right that should have gotten all of that just fine all right so we're getting close and we'll just do this manually for all these for all right so we get this Json file created now uh we should be able to go to our package Json and we're going to want to write a script say scaffolding is this so yeah what we're going to do is I don't think we need the EnV oh so that's the in and then the output would be right and the folder here is actually different so this is going to be slash uh Source slash shared and same thing for where the output's going to be all right so then looks like that worked it did not H uh and Justin develop thank you for the follow welcome e ah so it's not data Keys it's gen data keys that would be it so we should almost throw an error if you pass a command that doesn't exist I'll have to consider the right way to do that all right oh man so they're saying line 128 in that JS file [ __ ] that's the problem we didn't run another build for yep okay so now let me just do the hacking real quick give him Ana oh man I just added uh like my work calendar to my phone and it's giving me notifications about like meetings from last March so that's fun all right so version 55 is now published which means we just need to go into this file and update that dependency and now this next like data Keys command should totally work so yeah the main reason we have that entire like I got to publish the the package and all that flow is I've built my own like framework wrapping around a thing called architect and that basically allows me to get up and running a bit quicker and have oh well that's a problem uh nothing's ever easy um so yeah we got another bug to fix but basically having that framework allows me to spin up like new projects pretty quickly and still have like the niceties of the omm that I built the niceties of uh some of the other abstractions around like generating an open API dock things like that but yeah so we got another bug here and they're saying that's on on line 143 of CJs no matter what we always need a partition and a sort key so technically we don't need it there but we will need it here and I guess we should have been able oh why am I editing that file well let's just double check and see that it runs it does all right so let me go fix it but there's a flaw in what we're doing here yeah and our type isn't going to be able to know that so yeah here we would just want to do and yeah let's switch over to some synth wave here in a second we'll let the song finish but I keep forgetting to do that do it from this folder uh but one thing all right so we're going to add that all right we'll do just one more publish just so I don't forget about it later all right there we go talk to the hand right so that should run generating data key maps sure probably don't need those two logs but we'll call it good yeah I should have wrapped them around a uh like a debug call instead but yeah we'll call that fine so that should have been generated now and we see yeah data keys. TS oh all right so number one yeah we're getting like a little too much spacing on these but it's not the end of the world the big problem is that that doesn't seem to be there anymore am I not exporting that from data wrapper I think that is actually just uh the typescript stuff not being right yeah yes that was just uh the language server not being up to date for all right so we don't get any other errors in there we can ignore some of those types and yeah let's just clean it up oh and yeah we're not using DB keys there anymore so now middleware yeah we're going to have a bunch of issues there oh but I messed that up in the middleware okay so data wrapper looks good now but middleware breaks yeah no we did break some other stuff and yeah hold on no no let's just switch it to synth wve yeah the lyrics are getting distracting so yeah wrapped data store for e e so this is the trick part here this part breaks on us we can just make this an array of strings for now and then we'll think about how to make that like a better type for for e e okay so create data wrapper now just takes the table which is a data store the doc client e yeah okay so here's the big thing websocket tables and and the other rapper here right they no longer take in the key manager Ah that's why we do that so we do the gets that we can pass the T then yeah we want to do for Ah that's in the websocket folder okay so that one will be fine uh wrapped data store no longer exists for interesting e I'm not sure what that's trying to tell us oh and you know what hold on for all right so that's still not going to be the fix okay so that ends up working there perfect okay cool ah yeah so now this middleware and yeah this becomes the tricky bit and yeah what up sford how you doing for shy te nice never been much of a a chai tea drinker myself but it is delicious it's just uh it tends to be really sweet see me double check one thing here oh and yeah yeah yeah um anyone out there there is uh two pretty large bounties for some open source work if you want to do it could be cool right let's look at it together all right so create an open source guey for Gollum in typescript so build a typescript application that allows graphical management of Gollum which is some Cloud thing right serverless stuff supporting all the apis that gum supports requires knowledge of typescript front ends cross browser Dev user experience user interface design and developer experience serve the application as part of the single executable build of gum which bundles everything together in a single executable this may require a tiny amount of rust knowledge but you can probably get by with Google and co-pilot right it must support component management worker management API management plug-in management for Gollum Cloud $10,000 so depending on where you live that could be a significant Bounty even for someone living here in the US 10K is basically a a month of work at 100 20K a year that's pre-tax of course but you know if you live somewhere that's cheaper than the US it could definitely be worthwhile right where like the US dollar goes further because I do think this is about a month worth of work personally if you consider it like a hackathon it's not a bad deal right because hackathons tend to run for about a month you know there's not even a guarantee you get it same idea here but probably less people going for it as well then the other one is also Gollum Cloud 10 grand as well and this one is right develop an open API export for their current API def definitions so this one's definitely like more rust and backend focused but also maybe something kind of fun for anyone who wants to get their feet wet with some Rust and this is on Al Gora yes absolutely no not algorand so they currently have 25 Grand in bounties at the moment could be kind of cool could be kind of fun could especially be a good way you know to if you do well and Gollum likes it it could be a nice way of getting paid for your effort but also treating it as like an interview process for Gollum Club I don't think I saw any place that they were hiring yeah not currently but I don't know could be a thing damn this person got 15K yeah bounty to hire right so for some of these things if you solve the bounty you can get paid for your work and then also it could be basically like your interview process for the company looks like one person actually completed it bunch of people started some like Works in progress and didn't do [ __ ] and this person actually like finished it so yeah they started on October 27th and if we look at it November 12th is when they finished so they put in basically two weeks of work and got 15 grand it's not bad it's not bad but just keep in mind you know there's no guarantee that completing it will actually get you the Bounty so that is like the danger of doing any of these things like you saw 20 people attempted it one person actually got there so pretty pretty cool though all right so continuing along with what we're doing but yeah I thought that was like a really cool thing like I really like what those guys are doing with uh Alora yep yeah the problem is that it's like a large large project uh and nater one second there you go yeah yeah right like learning some js fundamentals definitely worthwhile but not spending too much time on them and then yeah now that you're learning react you can take some of those patterns and ideas and then if you want to go like fully native like you know just pure JavaScript again you've got some like best practices about how to structure an application better right so yeah I'm glad you're liking it I just think like trying to build an application with pure JavaScript without learning a framework is going to be really tough there are people who can do it but you're going to be like learning a lot of like bad habits along the way before you realize what like the good habits might be and like the biggest like habit that I think is good is declarative UI right based upon rendering whatever data you pass it rather than doing like direct Dom manipulation a lot of people get into the direct Dom manipulation thing and then learning proper declarative uh like Dom rendering becom trickier because they already like set up a bad mental model so yeah yeah no reasonator I'm glad you're enjoying it like yeah I I think building applications is oh man it's just so much fun it's just so much fun uh gleam seems cool I haven't touched it yet I know a bunch of people really like it it does strike me as the kind of thing where You' probably end up needing to like write a bunch of stuff yourself because the ecosystem might not not have it yet and the fact that you can't use existing Elixir stuff with it I think is like a a flaw right there may be some way of doing that but it's probably not as straightforward as it should be that's what I think gleam should be focusing on a like a gleam Elixir interrupt but you know whatever dude building it can do whatever the hell he wants LS pfold I think is his name the goal for gleam is to be a typ safe beam language but with a a nicer syntax than erlang that's about all I know about it honestly um I would highly recommend uh if you like gleam I believe DM mroy does gleam quite a bit on his stream and if not at least next time he's live you can ask him he loves talking about gleam at least but I haven't touched it yet so I I have no like solid opinion other than I like the idea of it n you can like nah you can learn typescript along with react for sure you don't need to get deep into typescript typescript and JavaScript are the same thing typescript is just JavaScript with types right avoiding the any keyword and being strict about that is a great way to get started but in one of the talks I proposed for react Miami one of the things I'm going to get at is the reason that typescript one is it allowed you to do the incremental adoption easier so if you wanted to adopt typescript into a JS codebase you could just allow implicit any and a whole bunch of stuff and then it's just JavaScript and then you can add types as you need to the power of that is that you were able to incrementally adopt typescript into your code base because it's still just JavaScript with types the big reason that that is not the case with flow is that writing flow is much more strict especially with the later versions like you have to type everything they they really just don't like the uh implicit any type stuff so yeah I would say yeah you don't need to know anything about typescript you can learn it as you go and that's the main reason it won I think compared to flow and basically any other like strongly typed compile tojs language it's whyatt one compared to reason reason wants you to be very very strict about your imported libraries and you have to have typings for it if there isn't a typing like you got to build it yourself and that's one of the things we had to do at get Kraken right in their flow code they had to actually like manually type a bunch of stuff right of the libraries that they were bringing in because of that it actually turns out that the types they created do not onetoone align with the actual types that the libraries now export and that causes some issues because of how we're using it it doesn't one to one line up with what exists so that can be a problem yeah typ hero is a really great idea um it was built by uh trash Dev and a bunch of uh like a few other people I don't know about a bunch I'm not sure but yeah type hero is a really great idea it's like you know it's like leak code or whatever but strictly for typt script so very very cool like very cool they even have their own like Advent of TS right so instead of Advent of code you could do Advent of typescript pretty cool very smart I just I don't know anytime you're getting deep enough into some of this BS is probably just not worth your time like you saw Rec like I don't know if you were here the past few days probably not but if you were here the past few days I was really like fighting the type system at which point it just wasn't worth it and I decided to just use a script to build out the typing stuff for me do I think everyone should follow the well [ __ ] the types are too hard let me just build a script to scaffold things for me probably not but it works for me works for what I'm doing I'm a big fan of scripting to like get around things that [Music] suck yeah oh you didn't get to see the solution well yeah let me uh get to it so basically in our CLI tool what we ended up doing was pivoting uh from doing like all those like type Shenanigans into having a Json file that we just you know crawl with this script tool and then we just do a bunch of like string templating logic to export these like actually like strongly typed key map values per table and then it's just you know like users key map dot get partition key right becomes very simple at that point so the put ends up looking like let me find one for you right we take in this data keys. Json and then we end up getting datak keys. TS by just generating stuff and crawling that Json file and then we get get partition key and now we have a strongly typed key there now as far as like the all stuff goes less important but let's look more at like the businesses keymap so getting a business is just you know it's strongly typed in that we need a business ID and you pass that string and then we create the string programmatically that way this function then needs to be way less typescript safe right like way less type safe because we're doing the type safety at this level instead which is like the like in many cases libraries don't need 100% type safety the only type safety they really need is the API that they expose to the external user so some of the most complicated libraries out there will actually use any internally just to make their job easier as a Dev as long as the external API that devs need to work with is fully type safe then you're good to go right like use the type system until it gets in your way and then just be productive right let tests fill in for some of that other stuff even then hex like how easy do you think it is to read one of those giant types that has like the turnery type system I forget what they call it right like dependent types or whatever how easy it is a new Dev like going to step into something like that too it's going to suck right like it's really going to suck you feel like this is harder to manage why right like feelings are hard to argue can you actually explain that feeling or is it just some hunch that you have because at the end of the day all I'm maintaining is a Json file and I'm building out functions that take in strong types and spit out a string so how is that hard to manage for the end user this is super easy to manage it's just a script you run and you get some functions that you call done now we're in the process of still like solving some of the other type issues because I forgot to mess with my middleware so yeah I do have to keep refactoring some other stuff but as far as me as the dev it's going to end up looking like hold on oh I deleted that file though didn't I yeah H I don't have that file anymore to show it right the usage file code review anti-patterns and that's all they use hold on I mean a belief and a feeling are the same thing right they're ephemeral they're unprovable because there's something you have that you can't justify right they they require Faith faith-based arguments don't work in code faith-based arguments don't work in science all right so hold on this sounds fun code view antipattern the death of a thousand round trips The Ransom note the double team the guessing game the priority inversion the late breaking design review The Catch 22 the flip-flop but seriously folks Authority and gatekeeping code review right like what does it mean so let's dig into it blah blah blah so the death of a thousand round trips you start reading the code as soon as you see a nitpick you add a review comment pointing it out then you stop reading theay developer conscientiously fixes your nitpick and submits a revised patch you start reading that version you spot another nitpick independent of the first you could perfectly well have mentioned it the first time but you didn't because you didn't get that far so you point that out and stop reading again yep death of a thousand round trips yeah it's like death of a Thousand Cuts sure yep get every nitpick out of the way as soon as humanly possible right you should as a reviewer you should give yourself one round of nitpicks if they introduce a new nitpick with their new code fixing one of them that's a different thing but if it's something that existed on the first time that you submitted one nitpick of like this variable could be named better sorry no too bad you had one round one chance to do it you didn't now it's about actual code review not just nitpicks but I will say there are certain times where you have a nitpick someone makes a change and in that change they made well yeah come on we could do that better too like come on man so that's a little trickier and yeah it's especially worse if you're in a different time zone because then it gets that like 24-hour round trip instead and that PR never gets merged all of a sudden so then there's another The Ransom note this particular patch seems especially important to the developer who submitted it maybe they said that outright or as part of their argument trying to persuade you to take the patch or maybe you just read between the lines but this patch isn't especially vital to you so you're in a position of power now you can hold the change they need hostage until they do lots of tangental related work which doesn't really need to be in the same commit but which is important to you if you ever want to see your beloved patch again that's good yeah sex thank you for the follow I appreciate that now these are good all right so the double team one patch two reviewers every time one of you Demands a change the developer obediently makes it and now the other one has a complaint instead so keep taking turns to make incompatible demands but always direct your comments at the developer avoid arguing directly with each other on the review thread and don't acknowledge any suggestion from the patch author that the two of you should talk to each other and come to a consensus about what the patch is supposed to look like see how many times you can pingpong the developer back and forth before they give up in a real emergency if you can't find an accomplice you can try contradicting your own previous review comments but normally someone will notice it works better with two reviewers what's my opinion on CD um I think CDs were higher quality than most MP3s we listen to for sure I know that's not what you're asking but I don't know do you have a more specific question about CD like continuous delivery and yes it's fundamentally true CDs are a higher Fidelity than MP3s like 100% I would argue that CDs are actually higher Fidelity than records as well but there are record people out out there in the world that would argue that point because well no the the digital like audio wave is stepped because it's digital whereas like the analog is smooth yeah but the ability for the needle to adjust based upon that smoothness actually changes over time as the record gets played so the more you play that record the more equality you actually lose from it right but some people fun just like the crackly sound of a record more than they like the actual digital Fidelity of a CD but for a record to have the same actual Fidelity of a CD it would have to be so much more dense on like the grooves in the record than any record ever produced right the record would have to be thicker than any record ever produced to actually get that Fidelity between the top and the bottom of the wave but I digress I've had this argument with a family member before you'll never convince a like a a record supremacist that CDs are actually higher Fidelity but it's true all right so uh the guessing game there's a problem and lots of different possible ways to solve it I know right four-wheel it's tough It's like really rough it's really rough yeah record supremacists they fully exist oh they're out there I have an uncle who's definitely won here's the biggest thing for a wheel once you beat them on the actual like physical argument the only thing they have to depend on is the feeling of it well I just like the sound of a record better well cool you can like it better but you were arguing about actual Fidelity don't move those goalposts the moment you beat them in an an actual like physics based argument about that Fidelity they will just fall back to well I just like it better we keep liking it better but it's still not technically as High Fidelity but now that I brought it back up again you're totally going to bring it up at the next Thanksgiving I get it I get it that's science I think here's the best part you both can agree on the fact that MP3s are the worst right MP3s are worse than a CD and a record so you're good to go right my favorite thing about MP3s though is that as you like keep them on somewhat faulty storage mechanisms right like if you keep m p3s on a hard drive there will occasionally be a sector that goes bad on that drive and you'll actually introduce degradation to that digital file so MP3s actually have more of a chance of being somewhat like a record where that MP3 is its own unique MP3 right there will be subtle differences between record to record based upon how how much it was played same thing with MP3s based upon how often it was copied from like you know medium that could degrade to medium that could degrade MP3s actually do have a similar like they degrade overtime philosophy based upon the storage medium they were on I have MP3s that have gotten like losser based upon bad sectors on a hard drive it's kind of crazy and I've even chatted with someone about it but but it's digital yeah yeah yeah but how are you storing it is there a chance for that digital medium to degrade absolutely yeah it's why the old um right like the the mini hard drive based uh iPods would also introduce like subtle degradation as well it's kind of crazy yeah I'm even willing to bet that like solar rays could introduce bit shifting or bit flipping on ssds yep yeah yeah yeah it's kind of a fun conversation to talk about like Digital Data degradation because a lot of people treat it as like no it's just ones and zeros it can't degrade that's only true if the medium it's on can never degrade and pretty much no medium is degrade proof right which is kind of crazy like even ssds can degrade just due to like solar rays and bit flipping so yeah like yeah really fun like really interesting to think about this stuff none of our data is actually forever yeah there's an entire generation of people that downloaded MP3s from Kaza that had the exact same like hiccup in the audio file and if you listen to that file enough when you hear someone else play it you could actually spot the fact that they downloaded it from the same place you did that's actually like really interesting to me too like so much so that when you hear it without that hiccup you're like oh oh that's what the song sounds like oh crazy Cana form oh there you go yeah digital signatures like that like very very interesting very fun like I don't know sorry I nerd out about this stuff but yeah hold on so real quick someone asked a decent question about continuous delivery uh you're struggling to convince the head of kway that doing everything manual including releases Etc isn't great he likes to do he likes the whole manual release process to staging EnV to test it Etc and convincing him other otherwise is an uphill battle I I mean there's a quote there's a quote um convincing someone of something that goes against their economic reason for being is impossible right you're basically trying to talk that dude out of his job you're never going to win that argument I'm sorry like yeah like what is the quote hold on H it's a lot of like how to BS now there's like a famous quote around this that I'm not going to be able to find because I don't remember the exact keywords he's basically told you that it's job security as well to be fair so at least he's honest you know how you win that argument and yeah it's like H Mendoza says you go to a manager above both of you and you convince them how much they could save on QA all of these arguments are one economically if you can show the economics of it like hey you're going to save 400 Grand a year by not paying these [ __ ] QA people you can keep one of them on un sure like dude you tell a company they can save 400 Grand a year they're going to do it that's actually like a really easy argument yeah like just follow the money prove out the money case for it I think the best case scenario is you convince the QA that his job's not going away it's just it's just going to change right now he's no longer a bottleneck for release he can simply be there testing stuff as it's released and just grab bug tickets to like add and it's an asynchronous process instead of a synchronous process that might be the move just try and convince him that no longer like when we do a release is he in a time crunch no if there's a problem he can just report it when he finds the problem no problem like you know like automate the parts that can be automated and then the parts that can't which he is totally responsible for well those just happen after the fact no big deal he just creates a ticket and the problem gets solved no big deal right so that might be your your way of convincing him he is no longer a synchronous bottleneck he can do his work asynchronously and enjoy his time instead of being stressed uh yeah so Random that's also it like it's easier to ask forgiveness than permission that is a somewhat loaded statement and problematic when dealing on like a personal level with people but as far as like business processes go it's 100% accurate uh absolutely soy I would say you don't want to be a tester QA cuz that job is going away right you're basically in the Iron Age saying like well can I still still get a job manipulating bronze you can but it's a bad idea hold on this is also kind of my favorite joke about QA uh QA engineer joke here it is right a QA engineer walks into a bar and orders a beer she orders two beers orders zero beers orders negative 1 beers orders a lizard orders a null pointer and tries to leave without paying satisfied she declares the bar ready for business the first customer comes in and orders a beer they finish their drink and then ask where the bathroom is and the bar explodes right because no matter how good you think you are as a QA engineer you're going to have gaps was ordering negative one beers actually a realistic goal probably not but a QA engineer thinks it is it's actually worthwhile to double check it but should it be part of their like Playbook to do so no that's something that you should have a unit test for QA should be testing the happy path anything outside of the happy path should be part of an automated test right Q QA should be validating that you can collect money from the user that they can do the number one thing that you expect them to do if QA is focusing on all sorts of weird edge cases those edge cases should have been like automated in some way now granted like no no this is actually smart but that QA engineer if they were a better engineer would have helped you write the tests to cover these things and not make it part of their playbook right like the QA Engineers Playbook should be the happy path I'm willing to bet uh random who asked the question your QA engineer is not only checking the happy path that's the thing there you go right they should be writing those tests if they're a really good QA engineer they should be capable what could you do to help them hop on a pair programming call and help them write the tests [ __ ] it man but they should not be manually doing all of these steps at all at all these should be unit tests or at the bare minimum integration tests or at the bare minimum end to end tests that a human being doesn't have to do anymore right the human being doing QA should validate the happy path the thing that gets you paid the thing that the customer does that they pay you for not edge cases so I would say two beers and zero beers valid those should be part of their playbook these things and maybe these three should be automated this one could be part of the Playbook too yeah I've never been at a company where QA was the actual bottleneck you know what the actual bottleneck was in pretty much every company I've ever been at code review yep code review and I am just as guilty I'm not pointing fingers like no I like I set aside maybe an hour a day for code review if your [ __ ] wasn't ready during that hour a day I'm going to do it well I'll do it tomorrow because I can't spend all day doing code review or I'm not going to ship any of my own code code review is everybody's bot neck th% it always will be because you can't just have one person whose job is only to do code review that doesn't make any sense right that makes zero sense so instead every single Dev on the project is juggling whether to review code as a priority or whether to ship code of their own as a priority and every single company in the world would prefer that you ship code as a priority there's never going to be a product team or a manager or hire you know like a you know leadership team that would rather have you be focused on code review as your core function they just don't no definitely not so yeah that [ __ ] sucks um I don't know how to solve it right I do work at a company called get Kraken where I do think we have some features that could help obviate or at least like alleviate some of these problems from the code review perspective you can get notified you can see a dashboard of all the things that you need to review blah blah blah like right there in your editor in in git lens right in VSS code Etc but also yes I get paid by this company so I'm kind of paid to say these things I still actually believe them oh really oh that's cool I mean honestly random I'd love to chat more about this I'd love to hear more about it in chat with you if you want to hop on a call at any point just send me a message over on disc ORD I do have some messages like that people have sent that I need to like follow up on cuz I'm slacking on that but I'll happily like schedule a client call to understand how you're using it and what we could do better it's my job I'm a developer Advocate like yeah come on of course they're awesome I work there no no no right I I'm I'm not paid to shill I'm paid to actually like you know be honest we got product marketing people that are paid to shill I'm paid to actually like speak on behalf of developers I'm here to bring your opinions about what sucks about it back to the company if there's stuff that you don't like about get Kraken I'm the person to talk to yes yes it does require signing in and that is a bummer now I totally understand there GS Vault and that is a common complaint and I will bring it back to them yet again but at the end of the day we're owned by private equity and private Equity requires like value on their investment I know and I get that if your code is open source you should be able to use most of our features for free anyway I think they've made that worse though too I would simply say it should be pretty easy to convince your company that $50 a year is worth it remote file system access is painfully slow ooh okay so Space Case could we schedule a call for that yeah I think we've made the open source experience worse and that is something that I've tried to push back on but I just don't have enough say I'm sorry I I'm oh I am so so so trying to convince them that we need to do what G Butler does or I think it's get Butler it's some other competing tool basically establishing a freeo use tier for open- Source contributors or startups that are small enough I would [ __ ] love it if we did that [ __ ] but I can only push so hard so often it's like an every three months thing where I just point out like hey our biggest competitor let's small startups and open source companies use their [ __ ] for free we should do it too so yeah I hear you GS volt yep that's one of the biggest things I'm trying to push back on uh GS VA you're talking about like um the uh commit grath I believe yeah the line diagram kind of thing yeah yeah yeah for sure the commit graph is one of our big selling points I will point out that VSS code has their own builtin commit graph the only thing VSS codes like native commit graph doesn&am