How'd you like to listen to dot NetRocks with no ads? Easy? Become a patron for just five dollars a month. You get access to a private RSS feed where all the shows have no ads. Twenty dollars a month, we'll get you that and a special dot NetRocks patron mug. Sign up now at Patreon dot dot NetRocks dot com.
Hey Carl and Richard here with your twenty twenty four DC schedule.
We'll be at as many NDC conferences as possible this year, and you should consider attending no matter what. The Copenhagen Developers Festival happens August twenty sixth through the thirtieth. Tickets at Cphdevfest dot com.
Ndcporto is happening October fourteenth through the eighteenth. Tickets at Ndcporto dot com.
We'll see you there, we hope. Welcome back. It's dot net Rox. I'm Carl Franklin and Amateur Campbell and we're here with Mark Rendell. We're going to introduce him in a minute, but first I want to check him in by Buddy Richard. You ready for dev Intersection Well, and this show publishes.
We will be most of the way through the show, I think we'll be close to doing like the closing session and stuff as we usually do. Right, Yeah, yeah, yeah, No, we got our got our act together, got a new version of a generative AI show, which has been an adventure to put together. Boy boy, there are a lot of AI conferences right now. Yeah, I know it's little nuts.
But still dev intersection with all your favorite Microsoft people and YE speakers.
The Hanselman's and the Hunters and all that good stuff.
Yeah.
The Scott's, Yeah, the major Scott and the lesser Scott's.
They're all vps. Now I mean this, there's one Guthrie but he's not there. But you know he's on sabbatical. What's up with that? Yeah?
What is that with that? What does Scott Guphery do on sabbatical? Does he read tech manuals? Does he create a few more patents? No, he's got teenage boys. I think he's got other things to worry about. Yeah, that's good. All right, Well let's jump into better note of framework. Roll the crazy music.
Awesome? Alight, do you go?
So this was our clickbait story on security this week.
But it's not clickbait if it's true.
Yeah, I know, but that's what we call him.
We called it.
The last story is always the one people want to hear about the most, you know, so we let them slog through all the other stories and then we spring it on them. So here it is. You're you're not going to believe this. Researchers find sequel injection to bypass Airport TSA security checks.
Nice.
So what this means is there is a vulnerability where somebody could use SEQL injection when they bought a ticket, and that ticket could give them status as crew, right, and then they could walk right on, go right up into the you know, the bulkhead seats or whatever, sit down and enjoy all of the crew you know amenities.
Not that there's not many of crew amenities, but yeah, you're definitely gonna have access to more seats.
Yeah.
The bigger one here is that you know, you can go to the crew line at the TSA. Yeah, yeah, and they that's a very cursorly view. And you don't have to be dressed as crew, I mean, you don't have to be working on that flight, right, but if you are crew, you get a special pass and you get all sorts of amenities. Yeah, interesting. Yeah, they normally check ID, So I wonder if you actually try to do that, if if when you got to the line, they're like, where's your ID?
Now that's just you know, raining on my parade here, Richard.
I still love that that, you know, sequel injection attacks.
Yeah, sequel injection. I thought we were done with this.
Nah.
No, still like it's number one or number two in the O ausp tense.
Oh of course, yeah, yeah, still still it was it's like sixteen years ago we learned about this.
Yeah, yeah, it's been it's been decades, and it's just a you know, sanitize your inputs. Bobby Tables is out there.
And Bobby Tables you got to watch out for him. All right, That's what I got. Who's talking to us today?
Richard grabbed a comment off a show eighteen eighty eight, the one we did with Steve Smith talking about modular monoliths, and I know we're going to talk about simplicity with Mark, So I thought this conversation just tied in so well with that whole problem space, because that's the show that Smith did focused on, you know, simplifying a monolith, tearing out pieces to optimize them. That kind of thing that whole idea that you don't have to go totally micro services.
You can just do what you need to do. And Mark had this comedy. He said it was a different Mark, not our gas card. He said. From an application level, I agree with Steve's point of view that we to start with a monolith and then split off whatever needs to be spun into its own services as the need arises. From a hosting point of view, though the reality is most businesses unfortunately, when you start down one technical direction, regardless of what it is, you're likely to keep going
down that forever because temporary solutions become permanently yep. For example, moving from something like app service to Kubernetes is a huge jump and it'll be hard to get business buy
in for what the need arises from it. Yeah, and I found you can use something like Kubernetes as a service from the onset, but with very minimal setup thanks to manage the services like as your Kubernetes services, which gives you the flexibility to about resilience and scalability and everything else as the need arises without having to overhaul
the entire system architecture. And I appreciate the thinking Mark, because my workaround would be to stay in app service, but then to spin off the problem service as a logic app right right, that just you know, make it a servilest component. It just gets called the same and you can keep it in the security context. But I'm not going to argue with you about moving from app service to Kubernetes. That's a huge lift. But that's the all micro services mantras to do that huge lift, and
I don't think people want to do it. So the fact that we can spin off bits into servilest components and deal with their performance issues or their behavior issues separately is the simple workaround rather than looking at the big wall of climbing up Kubernetties.
Yeah, that's why we have modular monoliths, so you can break off the pieces and need broken off.
Right, and you could break them off in less painful ways. Yeah, I mean, you could even spin it up as a separate app service, but I wouldn't.
Nope, I wouldn't either.
So Mark, thank you so much for your comment, and a copy of music co Buy is on its way to you. And if you'd like a copy of music Code, I write a comment on the website at dot at rocks dot com or on the facebooks. We publish every show there, and if you comment there and everyone on the show, we'll send you copy of music.
To go buy. And you can follow us on x Twitter if you want. That's we've been there for years. But the cool kids are hanging out at mastadon. I'm at Carl Franklin at tech Hubs Social.
And I'm Rich Campbell at macedon dot social.
Send us a tute. It's another way that you can get a copy of music to code buy. Absolutely all right, so let's bring back Rendel Mark Rendel. He is one of our favorite guests. He's the founder of Rendall Labs, which provides consulting services and workshops to dot net development teams across all industries. His particular obsessions are API design
and development, performance observability, and code based modernization. He also uses skills acquired during a few years as a professional stand up comic to deliver entertaining and informative talks at conferences around the world, and recently learned to play bass so he could join tech parity band The Line Breakers.
Hey dude, hi man.
Congratulations, Welcome to the world of music for no money.
Thank you, thank you so much. I am loving it.
It's been a while, It has been a while.
You did the parity band thing for a while there, Carl. But did you actually get into trouble with that? Was that back before parody was like accept the use? Oh okay, well yeah, here's the story.
So Don Box and Ted Pattison and George Bullock and a couple other people started this band called ban On the Runtime, and band On the Runtime was parodies of C sharp to well known music, right, so fifty ways to send a message, you know, that kind of stuff, right, And so the Microsoft lawyers had a problem with it because parody law says you can parody a song by you know, changing the words like weird al does like because I'm fat, I'm fat ham on hole week, right,
But you can't parody something like C sharp using a song that's not covered by parody law. So you can't change the words to parody something else. We weren't actually poking fun at the songs. We're using the songs to poke fun at dot net. And they, the lawyers kind of put the kabash. It seems very sus well, but
the law is the law. So they basically went to Paul Simon, for example, and they said how much would it cost for the band to play you know this at a you know, Microsoft event, and he wanted like two hundred and fifty grand. So that's the law. Yeah, you have to get permission.
The law is subject to interpretation. I think those lawyers interpreted it very oddly.
Microsoft lawyers interpret things as badly as possible. Don't.
But you know, they are Microsoft. And if it was just me doing it, who cares? Because go ahead, I don't have two hundred and fifty thousand dollars, you know, but Microsoft, Yeah, might be a problem.
Yeah, And you know, all we are doing. We provide the entertainment conference parties and it's like there's a couple hundred people there and they're all drunk. And yeah, we don't parody any specic. The closest we have to parodying a specific technology I think is JavaScript, which was one in itself, and JavaScript isn't even a language, it's it's a well, it's a standard. I suppose I was going to say something meaner, but I went.
But no.
Line Breakers their bassist used to be Don Sign really, creator of F sharp and then he had to he got pulled into gitub next, the kind of Microsoft Research equivalent at gitub, but he was still at Microsoft Research and still sort of the benevolent dictator for life of f sharp and he just could not fit in conferences as well as all of that, and so then they were without bassist. And then I said to Dylan Beattie, how hard is bass because he's certainly got four strings?
Can it be?
And he went as, actually not, it's much much easier than guitar, And so he lent me one, and I took it home and I plugged in, plugged it into my computer with a sort of six point five to USB cable, and got Rocksmith Plus and tried playing a few things, and I think within five minutes of sitting down, I had learned to play seven nation army okay. And so it's not I mean, it's not an easy instrument.
And if there are any basis out there going oh no, not this again, it is definitely not an easy instrument. I've got one to like jazz and funk and disco and those sorts of things, but the learning curve is extremely shallow at first, so it's very very encouraging.
Yeah, right, you can play Smoke on the Water just by going but.
Club but chugging basslines.
However, of them bass you know, it's easier because you don't have chords and and all of that, but it is more difficult because to play bass really well you have to have a great tone by the way, yes, and also you have to have saw like bass player. You just to be able to play a couple of notes, isn't playing bass? You know, you really have to. It takes a while before before the funk or the saul comes out.
Yeah, and just getting in the pocket like right on the b and just giving tone to the drums.
But you don't have chords, but they don't have chords. That's your jack app story and.
Seeking nobody as.
There are a few very extraordinary bases that nine essential.
So what's on your mind? Randall?
What have you been up to?
I was at the Copenhagen Developer Festival last week. Rich was there and it was absolutely brilliant and I had a new locknote talk that I did which is called the Albatross Project. But they also asked me to do the free meet up the Tuesday night before the festival started. And I had a talk that i'd submitted for Copenhagen that they rejected because I just do lockmotes for in DC now apparently, but they said, you can do that talk at the meetup if you want. And the talk
was called how simple is as simple as possible? And this is both of these talks, they're related. So the talk is a sort of technical can you do this? Can you do that? How can you keep things simple? And then the Albatross project is the opposite of the Phoenix project nice, and it's about a development director who doesn't deliver anything for thirty five years because he keeps
trying to stay on the cutting edge. And yeah, and I just after I did the talk version of it, at which there was one slide where the slide popped up on the screen and somebody immediately got up and walked out.
He knew it was a right slide.
So I'm kind of like, yeah, this is working.
You don't think like his wife was in labor or you know.
Is no probably a React developer, Ora Angular or View because those were the three there were three JavaScript framework logos on the slide.
Anything where you're like, hey, where you go?
I hadn't said anything because we were actually having a technical issue at the time. And yeah, I mean it is possible it was the technical issue that you sent him out there, just kind of like, I'm not waiting around while these people turn the screen back on or whatever.
But yes, it's I kind of I feel like I owe people an apology to a certain extent, because in my forties I did a lot of bouncing around the world telling people how to contain the rize things and do micro services and event driven things and Kubernettes and all that sort of stuff. And now in my fifties,
I'm kind of wishing most people would stop. And so that's my mission at the moment, and that's I'm going to be trying to get this talk into as many things as possible next year and start pushing at it because we've just made everything, everything just so complicated in the name of efficiency, in the name of standardization.
Potential future scalability. That's the big one.
Yes, absolutely, But also you hear the term DX developer experience and developer experience, I feel like that should be an important thing in terms of your ID, your code editor,
your command line tooling, that sort of thing. But if you're getting to a point where the improved developer experience that you get from sort of CICD and microservice deployments and rollouts and all that sort of thing is actually getting in the way of delivering a good user experience because of the increased latencies that you get or the fifteen megabytes of JavaScript that a webpage has to download before it becomes interactive. Then there's some thing wrong with that.
We have had priorities completely the wrong way round.
Yeah.
One of the ways to keep that stuff fast is to keep it simple.
Absolutely, Yeah.
And so yeah, it's been my mantra, you know, do the simplest thing that works ever since I got started in software. That's been you know, stapled to my forehead.
Yeah. I wish I could say the same of myself, but at least, you know, I'm realizing it now and attempting to share some of that.
Well. I think part of this is we wanted to believe that if we, you know, embedded ourselves in this architecture, like if we understood all the pieces that gradually software would be easier in that space. I just don't feel like it ever got there.
No, No, I mean I think it did at a particular point. I think espn net mvc pretty simple, was relatively simple, and the idea of the pipeline. So what we kind of picked off with o win based on things like Python WSGI and Ruby's rack middleware, and then o win obviously kind of got rolled into asp net core and that was nice and simple. It was like, do this, then do this, then do this, then do this.
But then I think the problem is that as developers we think that if we're not changing the way we do things, if we're not evolving, then we are failing in some way. There's this whole thing about continuous learning, and so people kind of go, well, there's somebody says microservices, So I need to learn microservices. Now I know microservices. Why have I learned it if I'm not going to
use it? And then they end up breaking something with like two dozen end points down into twelve micro services and you're just like, no, I should.
Have learned that from doctor card, right, absolutely.
Yeah.
Well, I was like the argument that my service makes a lot of sense when the team is huge, that it creates a level of granularity that you can have a lot of people working in parallel and then have your fights over integration rather than fights over coding. Yeah, but most teams aren't that big.
No, they're really not. But I think a lot of the time, people who don't necessarily have a lot of different job experiences within the industry, they think that they're in a big department because there's like one hundred and twenty developers, all right, And then you're kind of like no, no, no, no, no, no, go and take a big Netflix or Uber or Microsoft or your average financial institution. I worked at Morgan Stanley and I think I was one of ten thousand developers
working at Morgan Stanley. So yeah, that and the other thing then is when you're at that scale and your team within that maybe a lot larger and a lot more different people on it. You may have a sort of architect leading the design of the system, and you have testers and DevOps people and all those kinds of things, and so what they call a micro service can well be bigger than the entire thing that other people are
trying to break down into micro services. And so yeah, there's this dissonance from people not actually realizing what scale they're at because it feels big to them.
They just haven't been exposed to the other things. What about the cloud native concept? Is this where you fall on that we should define it first.
Yes, So cloud native is essentially building apps that are designed to run in the cloud and leverage I think, you know, like the the term twelve factor apps has
kind of fallen out of use a lot. I think a Roku introduced that idea, and I think a Roku was one of the first cloud native style platforms, way before Kubernettes and containers and all that sort of thing came along, and it was build your app so that it consumes services that are provided for you and can be deployed by doing a Git push or whatever else. And you're not worried about infrastructure. You're not worried about VM size, CPU core count, all those sorts of things, because you're just.
No updating machines, just million new instances in rolling.
Across totally totally. And then you have things on top of that around reliability, which obviously you know poly in dot net. But when you're building for the cloud, you have to build for partition, You have to build for things that are likely to you know, things in the cloud go down for two seconds and then come back up again. Because somebody did a push and it just happened, and so you have to that. You know, you can't
just go, oh no, the database is gone. You have to go, well, I'll try again in a second, then in two seconds, then in four seconds. Obviously with that weird fuzzing thing that changes those numbers into something not quite surround.
But maybe that's what's happened is people misconstrued being cloud native with products like Kubernetes.
That's the thing cloud. So you have cloud Native, which is a concept and an idea and an abstract that you can aim for, and then you have the Cloud Native Computing Foundation, which is attempting to be for cloud
computing what the Linux Foundation is for Linux. And you know, Kubernettes is often called the operating system of the cloud, and there's going to be people going often called by whom, But so they have projects that go through kind of incubator and preview and then stable and ready for everybody to use. And so people feel like if they are going to be cloud native, then they should go down that like it's a checklist almost and go are we
running on Kubernetes, Are we using a service mesh? Are we using a document database, Are we using event sourcing and all these things that they have projects for, And you can be cloud native. You can build you could build I don't, please don't, but you could build an asp dot net web forms app that was cloud native
that covered all those things. Just with the various rules and all this sort of stuff, there's nothing stopping you from pushing an asp net where it forms app into an Azero app service in a VM or something or a VM or yeah.
And that just proves the point that just because you have checked off all the boxes doesn't mean using the rate technology for the job.
Well. Absolutely, And you haven't talked about the customer requirements anywhere along the path here.
Nobody's asking about that. The customer's kind of going. What I would really like is for the page to load in less than four hundred milliseconds, And which is the something limit. It's like, four hundred milliseconds is the point at which people go that with subality limits. And we've actually gotten to this point now. This is one of the conversations I was having with people where four hundred milliseconds feels unattainable. And have you guys watched the Spotify
documentary thing on Netflix playlist or something that's called. So it's about the guy who started Spotify. It's like six episodes and they're each done from a different perspective, a different person's point of view. But Daniel Lek, the guy who started Spotify, his sole focus to get it launched, in tech terms, was that when you click the play button on a song, it should start playing like immediately. And he was aiming for that four hundred millisecond point
and he got it. And that was in two thousand and one. And now we're here with an internet that is several orders of magnitude faster, We've got five G on our phones, we've got gigabit plus internet at affordable prices for a lot of people. And we're still saying
that's really tricky. We can do it in like one point four seconds if you're in a fast connection, And yeah, we've we've lost lost sight of the ball, and we need to examine ourselves as industry and say, well, what are we doing and why are we doing it?
And is there anything what are we doing for those one point four seconds? Exactly what are you up to?
And so yeah, and like I say, the talk covers this from infrastructure from back end, from front end, from programming language and all these different things, and yeah, frontend, it's it is honestly quite a simple answer. We are downloading a shedload of JavaScript which the browser then has to pass and jit and do all those sorts of things. And before we do all of that, people can't scroll
a blog. And you're like, it's a blog. The only thing on it that needs to be interactive is the comments and maybe a like button or a share button. Even the share buttons are just links. But no, I see people sort of writing blogs with nuts dot js.
Are you still bolished? And blazer server, Mark.
I really like blazer server for internal things, for where you have a very definite upper limit on the number of concurrent connections that you can have, and it's like in the thousands or maybe tens of thousands, and you can scale up for that and everything. Once you scale blazer server beyond a single instance, then obviously it becomes trickier.
So if you've got like messages and one person sends it on one instance, that now has to bounce through something like a reddish pubsub or something so that the other instances can pick it up. But it's still very simple. Actually, I've been playing with so blazerin dot net eight and you have the server side rendering, and so it will render pages on the server and it will return them, so you can have a site that works with search engines just as much as anything else, just like Razor pages.
And then it will download the Blazer web assembly runtime in the background, and it will download your natively compiled Blazer application in the background, and then silently without the user even noticing that's anything, that anything has happened, the next time they navigate, it won't make a request to the server. It will do it locally.
It's like magic, except that it doesn't work.
It's they'll get there.
I've done I've done tests and I recently did another one and we go from service side rendered directly to web assembly. There isn't any server rendering in between while web assembly is downloading. And that's been a problem ever since they released that, and it's still a problem. Well, but they'll get there.
I'm just going to keep my fingers crossed for for
dot netline. But no I did. I ran into some issues with it myself and have started looking at an alternative thing, which is HTMX, which is that X stand for so essentially this is a push back against things like React and Angular and so forth, where it's kind of what you actually want to do, what you actually want to achieve is when the user clicks something or interacts with something in some way, you want to make a call to a server and then you want to
make a change to the dom of your current page right and insert a div and with some stuff in it or something and so works, which is how Blazer
server works. And HTMX is kind of let's let's just lean into that as far as possible, and we'll just use HTML attributes on elements and we'll say, okay, so on this element, when the user clicks, we are going to make a request to this endpoint on our server, which is going to return a fragment of HTML, and then we're going to insert that HTML into the page at this point, or we're going to replace an existing
element with that HTML. And for I think like ninety percent of use cases, that actually covers most of the things you want to do. It will do your comment form at the bottom of your blog. It will do your like buttons and an updated view of how many likes you've got, and all those sorts of things, but without a huge JavaScript run time doing shadow dom things and observing signals. They seem to be going on about signals over in JavaScript land at the moment. That's very
fighting for them. And yeah, so.
That's sort of like the evolution of ajax, isn't it.
Yeah, yeah, it is, and it's kind of it's that the plumbing. I feel like it's a shame we had to go all the way round SPA to get back to the point where we're going. Here's a really sensible evolution of ajax.
Just refill a div tag for me, would you please? Right?
So I built a site when I was learning the base, and it's called fret Badger, which is partly because I like, it's not a mollusk. If I ever start another company, it's just going to be called not a Mollusk limited. What do you do? Definitely not a mollusk. So yeah, you can you can hire hijack the domain for me and then blackmail me. But no, So frat Badger. I wanted to learn the threat board, and I wanted to learn scales and modes for jazz and arpeggios and that sort of thing, and.
So I see chart behind you.
Yes, exactly. I've got the bas chart on the wall which has got all the modes and everything else. And it's basically just if you want to play up this scale, and you've got the major scale, the minor scale, pentatomics, blues, all these different things. And I was my base teacher was pushing me to learn music theory, and I got really into it because it's it's maths, it's rules, and it's logical, and so I thought, well, the best.
Way there's one of the many reasons that musicians and developers go work so well together.
Yeah, yeah, And I thought, so if I can encode all this information about music theory into C sharp and some data structures, then that would be a good way to learn it. And also it would be a fun site to build. And so it's just an asp net Core MVC app at the moment, but it does have this thing so you can have like you choose your freat ball and you say, okay, so I want the standard base threat board, standard tuning, and I want the C major scale, and it will show you the little
dots to show you where to put your fingers. But then when I was doing jazz, it was kind of it would be really good if I could have like four or five of them on the screen at once, and so I added in nothing. But every time you add a new threatboard, it reloads the entire page and then it scrolled back to the top. And so I was looking for a way to make it kind of more smoothly interactive. And so that's what I was looking at the Blazer server side plus WASM for. But as
Carl says, there are some issues. But then I saw HTMX and I'm kind of like, that is exactly what I want. And the beautiful thing is it works really really well with esp net core because we have partials and so we can just go, when this request comes in, just render this partial. So yeah, it's and it's that sort of thing. The key thing with HTMX though, is the entire library is fourteen point six killer bytes minified and chesipped, which is tiny.
It's less than your logo graphic button.
Radius is bigger than that.
Yeah, yeah, you know, it's like that thing where when Windows seven came out, the desktop wallpaper was took up more disk space than the whole of Windows three point one,
like and what is it doing that? So yeah, it's that, and you know it's They provide documentation and guidance on how to use Alpine js, which is again a very lightweight JavaScript framework, kind of like a miniature alternative to jQuery for the things that jQuery did that didn't then get built into JavaScript fifteen or twenty sixteen or whatever.
So if you do need, I have one thing where I wrote custom JavaScript, which is, once you've got your C major scale up on the threat board, if you click one of the c's, it will fade the others that aren't part of that arpeggio. And so it's kind of like going, so you do this, and then that's your second, and your third, and your fourth and so on. And that's twenty lines of chavascript, which isn't even worth
minifying at that point. I think it's actually still hard coded into the h into the layout dot c shtml and so yeah, just that little bits of JavaScript here and there to do a couple of things in the browser, and then HTMX for everything else. And it's it's working really well. I'm going to be pushing it live soon.
It's great. Before we take a break, I just want to mention that I had written I don't know four or five years ago this tool. I have a guitar that has a role in GK pickup GK three pickup, which is a split pickup, so it goes right in between the bridge and your first pickup of the guitar. And it has six individual pickups for each string and those go through a role in processor and then you can process each string individually. So another thing I can
do is map it to MIDI. So what I have is I have something that pulls up the MIDI from your guitar and figures out where your fingers are on the front board and shows it to you on the screen. Not only that, but then I put a service between it, so somebody could log in in Texas and I'm playing in my place and I can be teaching them, and as I play, the dots show up where my fingers are. Not only that, I got to chord charts and chord
names based on the music theory of it. So if you're interested in that codebase, I'll gladly send it to you.
I would love to see that. Yeah, that sounds amazing. So yeah, and I feel compelled to ask did you write that using events hourcing or DDD? Is how many.
It was very simple, all right, So we're going to take a quick break and we'll be right back with more with Mark Rendell. Stay tuned. Did you know that you can work with AWS directly from your ide Aws provides toolkits for visual Studio, Visual Studio code, and jet Brains rider Learn more at Aws dot Amazon dot com, slash net slash tools. You know it's common for business application to contain fifteen percent repetitive code just because of
metaprogramming limitations in the C Sharp language. Why write boilerplate manually when a machine could generate it for you? Enter Metaalama, the code generation and verification toolkit for C Sharp. Their C Sharp to c sharp template language is simply amazing logging caching memento observable. If it's repetitive, Metaalama can automate it. Visit Metalama dot net today and learn to automate your code patterns with their free edition. Remember it's Metaalama with
one L Meta la Ma dot Net. Attention dot net developers looking for the ultimate SDK to handle electronic document processing Meet TX text Control. Txtext Controls, your go to solution for seamless PDF generation, secure electronic signatures and efficient digital forms processing all within your dot net applications. Empower your products with robust document management capabilities, boost productivity, and deliver top notch solutions to your clients. Trusted by developers worldwide,
including me and Richard. Txtext Control is the SDK that makes a difference. Check look out demos dot textcontrol dot com for live online demos and see it in action. And we're back. I'm Carl Franklin. That's my friend Richard Campbell and Mark Rendel's here, and we're talking about simplicity. Simplicity. And by the way, I need to remind you that if you don't want to hear the ads, you can become a five dollars a month patron of dot net rocks and get an ad free feed. Go to Patreon
dot dot NetRocks dot com. Okay, so I guess you were talking about HTMX and.
So yeah, and HTMX is one example of simplification. People. Actually, I'm starting to see projects popping up that are pushing back against this complexity that we've created. And there are other examples as well.
There we are.
You don't need to run a coubonet. So the comment that you read at the start of the show from the other Mark and he were saying, how Aks and eks the sort of managed Kubernetes are super easy to run. They're really not they really they're easier.
They're easy. Err, it's only because you've been traumatized by trying to run Kubernetes by head.
I mean, yeah, never ever, ever try to run your own Kubernetes cluster, unless maybe you just want to build one with five Raspberry pies just for giggles. But no. So let's take a zero as an example, just because it's the one that I know best. But Microsoft support a version of Kubernetes I think for four versions for minor versions. So I think we're on like one point twenty one at the moment I get confused with version numbers, and so they will still be supporting twenty and nineteen
and eighteen. But as soon as twenty two comes along, if you're still on eighteen, you need to upgrade. And Kubernetes versions land three monthly or six monthly, I'm not sure, so.
It means every year then you're probably agreeing so.
But you are at least once a year going to have to upgrade your Kubernettes cluster. And if you do it, if you do the upgrade the minimum number of times possible, then you're doing a I mean, Kubernett's versioning is a whole thing anyway, because those are not minor versions, they are major versions. A lot of the time they remove APIs, they deprecate APIs, and yet somehow we're still on one
point something. And so if you want to minimize the number of times you have to do this, you're going to be going from like one dot eighteen to one dot twenty two. That's a lot of jump. That's a lot of things that in an enterprise environment, you're going to have people going over pages and pages and pages of release notes to make sure that you're not going to break anything. Or you upgrade every time the new version becomes available, which is going to take less time
per upgrade, but potentially more time overall. And then it's not enough to just be running in Kubernetes. You have to have your service mesh, you have to have your sidecars, you have to have your your sort of various load balances and all these kinds of things, and nobody's managing that for you. Microsoft doesn't Microft hasn't blessed a service mesh because they can't because people want different things. So it's it's simpler, but it is by no means simple.
And then you look at you type container into the creator new thing in the Azure portal, and actually the first thing that pops up is container app and that is simple. So that is build. You write your DOCA file or you know, in my case, right click in the solution explorer in jet Brain's rider and say add docer file and it just generates it for you with all the right best practices, and then do Docker build, and then do Docker push to your as your container registry,
and then just deploy the thing. And underneath it all it's running on a massive Kubernetes.
Cluster you want to have to own.
Yes, Microsoft are taking care of everything. They're taking care of the service smash, the upgrade cycle, the stability and resilience and all those sorts of things.
Well, and arguably, if you decide it isn't running well in container apps and want to move to your own manage infrastructure, you're already there exactly.
Yeah. And equally, now you can take a container and you can push that towards your app service where you used to push web things to, which is also very nice and much easier than saying copy locally and all.
These various things.
So yeah, you know, on that side, there's that simplicity. And again it's taken us a long time to get there.
And this is Microsoft figuring out what people actually willing to use, right like.
Yeah, yeah, and you know, you know, the reason you don't get straight from A to Z is because you can't see zed until you've gone through B, C, D, E, F and everything else. Yeah, and you don't know what it looks like. And it's very difficult to find out that something's a bad idea without trying it.
Yeah, and without well, and you always have the few folks they get it that they're willing to put the time in their their customer is tolerant of it, and they can do this complexity. But then a whole bunch of others just like Nope, not doing it. And then you you know that in the end they have to you know, Microsoft looks and goes, well, we're just not getting enough adoption here. What if we did this, like
they are just kind of hurling it at the wall. Yeah, by a bit, and eventually they stumble on to these approaches that are simple enough.
And you know it's the big thing has been serverless and still is serverless for a lot of the talks I'm seeing and that sort of thing, which is containers under the hood. It's containers under the hood. But the again I'm going to talk about a Zoo functions but equally applies to a ws lander and another Servilus platforms.
What you end up with then is vendor lock in, right, because your triggers you've got like blob stories triggers, and there's your Q triggers and service bus triggers and event hub triggers and all these different things, and so you're locking yourself into that. But with as your container apps, you can use Dapper, which is an abstraction over so you go, I need a cash or I need a pub sub and then that could be reddis it could be as your service bus, it could be men cash
d or Varnish, you know whatever. The different things are fusion cash and in your container description you just go, this needs a cash, give it a cash, and then you use the Dapper API, which will one just as well. You can use Dappa with containers running in amazons. D A p R Yes, d A p R naming definitely an issue at first, and Microsoft.
Well, Plus there is the other product called Dapper d A p p e R right, which is the.
Which is a wrapper around Adeo dot net.
Yeah, it's the kind of incredibly lightweight mapping from data readers to CEP objects.
I have nothing bad to say about either one, but they've got nothing to do with each other except when you say their names entirely unrelated.
But yeah, d a PR stands for Distributed Application platform something. But DAPPA is great, and it's I think dapper play is. I would put that in my simple column. It's kind of I need a cash. I need a way that I can say cash this data, now give it back, and I shouldn't have to care about out whether that's reddis or another solution or just memory. I should be able to say cash it, give it back, and then
worry about how that actually happens at deployment time. And again, you can use dapper in Kubernetes, but you don't have to. You can use dapper in other environments as well, and so, and I just want to push people in that direction.
I've well, and what I appreciate this is that you're still giving us access to all of this complexity without making it complicated.
Yeah.
Right, you're not saying no containers, You're saying containers that don't make you want to stab yourself.
Well, I mean containers. Containers are absolutely the simplest way of deploying an application. Sure, you know you need a Linux dependency like lib zip or something like that, Just don't make me edit apt get in the docer file and you've got it. Yeah, it's that, it's that original thing. If it works on my machine, I'm going to ship my machine. Yeah. And so yeah, containers I think are the simplest unit of deployment we've come up with yet. And it's just a matter of where do you run them?
And do you need to run your Kubernetes cluster.
Yeah, And you know we're we're not complaining about containers. We're complaining about the orchestrators totally.
Yeah. Management and the assumption that because sort of Facebook and Netflix and Uber are using something that it's the right solution for your company.
Right, You're not them.
You're not them, And they have armies of engineers whose only job is to maintain those that infrastructure.
And test new versions and roll them out and all of that sort of stuff, and all that stuff if you need it, but if you don't. At the same time, you don't want to cut yourself off from the possibility that you might become one of these things.
Oh yeah, yeah. But if everything's just docophiles and containers, and at some point you go, actually, you know, it would be a lot cheaper when you get to the point where you're spending so much on as you're container apps, that you could not only run your own cubernettes cluster, but employ somebody full time to keep an eye on it and do the upgrades and everything. That's the point at which you switch.
Certainly on the run as side of things. For me, I'm having conversations with folks who are like, no, we repatriated this because we already run a twenty four to seven knock and it was we were committed to doing these things in this workload. Now that we truly understood the shape of it in the cloud made more sense running at home. Yes, but that's a big lift, right, Like, it's not a small things.
Really, and the same thing goes I mean, I was complaining about people downloading fifteen megabytes of JavaScript or WASM or whatever to run something in the browser. There are
legitimate use cases for that. Adobe now have inside the network Photoshop and light Room as WASM compiled browser apps, right, So, and yeah, you know, if you want to edit photos and clean up red eye using AI algorithms, and you want to do it in your browser, then you need to download all that PLASM and that sort of thing, because you can't have somebody doing a round trip every time they make a change with a multi megabyte image file uncompressed.
Arguably, you're still stopping about you're still talking about the simplest case. In this case. The simplest case is stick the whole thing in a ASIM file and stuff for the download overhead.
It's that as simple as possible. And thinking about what that means for you for this project that you're working on today, it.
Feels like you're just pressing against the defiles that we're choosing complexity we don't need.
Yes, yeah, and I think that, and it's not I'm absolutely not blaming developers in isolation for this, because I think that the IT department, there were CTOs out there, there were CIOs and they're going to their own conferences and being sold this stuff by people with very very effective marketing machines, and then they come back and go weak.
And motivation is not involving keeping things simple.
Yeah, And they come back and they go, we have to adopt this because it's the next big thing, and not doing proper kind of cost benefit analysis and those things to work out whether it actually is the right thing.
Something just occurred to me about Blazer server. I remember having customers say they didn't like blazer server because if the connection goes down, like they get that what I call the transparent veil of death. Oh yeah, you know you have screens of death. This is a transparent so it turns slightly transparent or translucent translucent, and you know
you have to press the reload buttons. So apparently in that net nine that is becoming much less frequent, Like the whole automatic reconnection thing is just going to happen.
And I believe that they can make that acceptable to everybody, certainly to the real tech guys. So instead of it going slightly transparent, don't change the UI at all, right, just but it's frozen and it doesn't update and then just change the cursor to a spinning beach ball and everybody will be happy.
Everybody will know what's happening.
Yeah, it's like to say.
The UI and spin beach ball, and then when it comes back.
Why do they need to build a new view of that?
You know?
Yeah, that's the thing we trust you can you can be very subtle. I think it also indicates that possibly we will recover the veil. Doesn't give you a sense you're going to recover no veils, like call tech support. Yeah, what does this mean?
It's gone it's gone wrong.
We have pulled the sheet over the body drug.
But it does it makes sense in a context where you do have to hit control R or F five or whatever to refresh the page to get it to reload. But once they've got that reconnect and heal process working properly and smoothly, then so.
Anyway, I haven't checked it out in that NET nine, but I hear that it's a major feature upgrade of Blazer. There aren't very many.
Fewer veils is a good one. Yeah, that'll make people happier without a doubt.
Yeah.
The other thing that I talk about at the end of the talk is is process. How we went from kind of s S A d M and prints to and all the waterfall things. And then there was extreme programming. But extreme programming generally only works when you've got Kent Beck and Ward Cunningham and extream programmers and you know, some of the greatest programmers who've ever lived, all in a room together, going how are we going to build this? So you can't do that on your average team of
developers hired off of LinkedIn. And so then we got Agile, which was trying to make it so that we could where you get a scrum master and a product owner. And I've worked on multiple teams that called themselves agile and said they were doing scrum, that didn't have a scrum master, had a scrum master who was just a project manager anyway and wasn't really doing the scrum mastery things, and in an awful lot of cases, no product owner at all. Yeah, agile is is agile ish scrum butt.
And yeah scrum, but we don't do this except.
And then you see the scaled Agile framework for enterprise and the lean scaled Agile framework for enterprise, and the lean one is like for small to medium enterprises and it seems it's like the London Underground Tube map. It's just all the different things. And then when I'm sitting I'm sort of working on the new version of threat Badger, which is going to look exactly the same as the old version. I'm also working on another thing. I just
use can ban board. I have a backlog column and a ready to do column, and a doing column and a done column when it's just me because I considered the testing and everything part of that, but you because it will put uat and everything in there. And the real issue that came along with scrum was you have your backlog grooming sessions, which are like weekly regardless of
how many weeks you have in a sprint. Then you have your sprint planning session, and then you have your daily stand ups, which everyone tries to keep to fifteen minutes, but a lot of the time they run, you know, upwards of half an hour. And then you have your
your retrospective and your demo and your sprint review. And if you're doing two week sprints, you've got ten days and you're spending like twenty percent of that in the various meetings, which means you're not building anything you're not shipping anything, you're not delivering anything, and where's the value.
And if you are a small team of six or eight developers, and you are, if you're homogeneous, if you all know the domain and all the different technologies and all this sort of thing, then anyone can take any ticket from the ready to do column and say I'm doing this, and you can prioritize them just by moving them up and down. And you don't get into this thing and ah story points and T shirt sizes and all that stuff, and it's just kind of you don't
have to do all that stuff. You can just put a board up and put some stories on it and move them across and just keep shipping and delivering.
One step up from a notepad totally pretty much.
Yeah, you know.
When you mentioned the cam member in the first thing, it's just a way to visualize where we're stuck.
That's right.
That everybody can see that. You don't have to have a meeting. Nobody has to memorize, like you see it every day and you see it when it don.
Sticks yes, and you know people can kind of go, that was there yesterday, why is it still there today? I'll go and ask him rather than having a stand up. The idea that you set aside fifteen minutes in the morning, I think it appeals to developers is like, all right, so I have to talk to people for fifteen minutes, but then yeah, I get th headphones on, getting motivated to be a low Yeah, it's another it's that developer experience thing.
Again.
We don't like talking to each other, so just contain it to work things. But no, you should be getting.
Out of the way, have a couple of cups of coffee.
Show has just been about containers, right, It's just that.
Can contain the complexity. So yeah, So, like I say, I'm going to be doing that that talk. That's my my sort of tech talk for the next year at least. And oh, I would like to shout out Alex Russell. I think slightly late is his sort of handle on various media and everything. So and he worked on the Chrome browser for a long time, on the jamscript, run time v eight, that sort of thing. And he is like the little Dutch boy with his finger in the in the dam saying please please stop using all these
JavaScript frameworks and everything. And so I was some of the inspiration for the talk came from reading his blog posts as well, so shout out to him.
Great infrequently noted is his blood.
There we go, that's the one.
Yeah, yeah, I'll throw it, including the show notes. I've read it before. He's he is quite brilliant.
He is. He is a very very smart guy and we should listen to him.
Not wrong. Yeah, so you got this talk. Where are you going to be delivering that in the upcoming months?
So next year it's probably going to be at the NDC conferences, go to conferences possibly dev Ox. DeVos have started asking me to come and do things now because I've been doing all these you know how JavaScript happened and the Albatross project and programming as great as mistakes, so they're not tech specific and so I'm getting to go and do other conferences and meet entirely new audiences,
which is wonderful. One. I'm really enjoying it. And also my kids are at that age now where I can do more traveling and speaking engagements, and I'm really looking forward to getting into it.
Yeah, there's a couple of conferences in Sweden in Stockholm that happened every year. You have Sam and sweet Toob.
Yes, i will definitely try and get to dev some I've done that a couple of times. That's always a good conference. But yes, if anyone's listening to this and would like me to come to your your massive conference with three thousand attendees, but even if you would just like me to pop over and talk to a meetup or a user group or something, then get in touch and we'll see what we can do.
Check out the render lads link is on the show notes.
Just grab a busfair a sandwich in a hostel.
Yeah, that'll do me. Wake outch have base will travel.
Yeah, there you go.
Mark. It's always a pleasure talking to you. You always bring something interesting to the table and this is no difference.
So thank you, thank you for having me on again. It's it's really great talking with you guys.
All right, we'll talk to you next time. I'm dot net Rucks. Dot net Rocks is brought to you by Franklin's Net and produced by Pop Studios, a full service audio, video and post production facility located physically in New London, Connecticut, and of course in the cloud online at pwop dot com. Visit our website at d O t N E t R O c k S dot com for RSS feeds, downloads, mobile apps, comments, and access to the full archives going back to show number one, recorded in September two thousand
and two. And make sure you check out our sponsors. They keep us in business. Now go write some code. See you next time you got jed middle vansported the s