.Net Foundation Virtual User Group, 23 Sep 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Indy Xamarin User Group - The Best Way to Architect Your Xamarin.Forms App

Description

The title of this talk is a lie. There is no one single way to architect an app. If the app works well, is robust for future expansion, and maintainable - it's architected just fine.

You can mess things up though. And then everything is terrible.

So, tune in to this talk to find out about some ways you can go about architecting your Xamarin.Forms app so you don't hate yourself before development of the app is complete.

Presenter: Brandon Minnick, Senior Cloud Advocate, Microsoft

A

Oh now, we're live. Look at that. That's awesome, hey.

B

A

Hey how's, it going all right, hey long time, no talk, yeah yeah! It's been forever.

C

Yeah yeah, you were talking about moving and the funny thing is it's just 30 days ago. I didn't have any of this on my wall and into my home like in may of last year, and it wasn't until the last 90 days that I actually ended up with anything on my walls other than like white boards. Like that's the first thing that went up yeah, that's enough! Okay, it's cool! You know I can do this.

D

I know it's it's interesting in this age of uh elvis being at home and having them and being online because of that yeah. It's uh you almost feel that need to have cool decorations behind you, like um I don't know if you caught the ignite uh microsoft ignite conference keynote yesterday, but I didn't uh well satya nadella, the ceo of microsoft was speaking and yeah. He had shelves behind him with toys and trinkets on it and that just started a whole twitter rabbit hole of people trying to figure out. What's what's on the shelf and.

D

C

D

A baby yoda on there but yeah my walls are still blank. We just moved into this house a couple weeks ago to essentially move to the suburbs getting away from. We were living in san francisco and it's just not.

C

D

To live in a big city, especially living in an apartment uh during a global pandemic. So now we have a backyard and life's a just a little bit better.

C

Yeah, I think I know what you're hiding in the background there and that's the uh the.net maui preview.

D

uh I wish I can't wait, although because there's you have there's nothing there I'll catch it, I do have that queued up for today's talk, though I'll touch on that just a little bit. Did you see my presentation.

C

About that I did a whole thing about it. No, I didn't I'm excited about that. I said I think it went well, it was virtual, so no one would throw anything.

D

Yeah, you could say it went as well as you want people were. You got standing ovation.

C

A

That's right: everyone was really cool about it. It was perfect.

C

I wonder what the delay is on this I'm going to mute for a moment, so I can see what.

C

C

So it says, like seven, people are watching right now: hey xamarin. Does it.

D

Sweet uh 20 seconds.

A

Yeah, it's it's not that bad! Actually.

D

C

This is fun so yeah, the the whole maui thing, maybe I'll need I'll talk with you at some point, maybe offline. I got a whole bunch of cool bluetooth stuff. I want to consider doing I'm I'm curious what your thoughts are about, how best to approach that, from a cross-platform perspective on xamarin.

A

D

That's a good question: I've I've never done bluetooth myself! I.

A

D

It seems to always be painful whenever I hear anybody talking about it. Is that how it was for you.

C

um So the the the one time I did xamarin or probably bluetooth integration and a xamarin app was in like 2012., oh when it's probably back before bluetooth led you had to pay like a thousand dollars or whatever for those greater license.

C

Yeah back in the good old days, I was building an app that via bluetooth. This is all this is before ble over bluetooth, with inflate and deflate shocks.

C

And it was, it was really cool because I had to um figure out. Oh there's a question coming in about maui, um I had to figure out how to do this in a nice threaded way. Otherwise, you know you'd lose in the app crash.

C

It is all before camera and forms like xamarin forms would have made this so much easier, a ui perspective, but instead I had to like build the the zibs and nibs and all the other insanity. You know I was just like at least uh andrew at least um ios develop or harmony. At least threading was simple, with c sharp, like yeah, try and figure it out with like objective c.

C

No, uh no x, man like I'm like, oh, this is so much better his. So he because I can so easily handle all of this. um From a from a maui perspective, it is an evolution, you know it's an evolution, forms and and what the cool thing about it, I think, is if you, if you watch our previous presentation about maui, it covers kind of like how we got to maui it really how we went from the the mess that was microsoft.

C

Development of insert anything with an x at the end of it here and then all the ui approaches that microsoft's had over the years and finally seeing consolidation happening with net and then finally, seeing uh cross-platform being a thing with netcore.net standard and xamarin, and seeing all this stuff come together um and microsoft going hey, you know what dot net should be everywhere and we need to take this cross-platform stuff seriously, but we also need kind of, like we've consolidated the the libraries and the the api surface.

C

Let's start uh standardizing the the ui and maui gets us there. Now, it's it's not gonna be around until 2021. That's the the current estimate um with net six. My guess, my understanding is that uh with build this year in what november no dot net conf that night, conf in november, is when they're going to release.net 5 with we hope a an alpha or a beta version of maui. Okay,.

D

C

Would come along with net 6 in some form or another? Now my my hope here uh I haven't played with uno yet, um but I don't think uno is from microsoft. It's not me! If I'm wrong about that brandon, uh it seems like.

D

C

Approach um but I haven't, I can't talk about it only because I haven't played with it, but it's when I looked at it like a year ago I was like oh yeah, that's cool um and keep in mind. uh You don't have to switch to maui right now and first of all, it doesn't exist.

C

Maui is an extension of not extension up, but it's an evolution of xamarin forms. So you know you, you don't have to immediately move right. Well, I mean microsoft's, not a country, but yes, um I'm I'm joking there. Hopefully, hopefully everyone laughs, it doesn't take me too seriously, um but it'll be interesting. I think you know.

C

The ultimate thing would be if uh microsoft started really taking consolidation seriously on the ui front and got rid of uh win ui three and just said we're going to go all maui, you know and have a standard way of handling um ui in dotnetcore.net yeah. I think that would be great. I hope that's where it's going, but you know hey the cool thing. About.Net is it's so extensible.

C

I think. As long as you know, like the standard framework, is you know just consistent, which is where I think we've gotten to then you know you can do whatever you want. You know, xamarin doesn't require you to use. Xamarin forms, that's not a requirement.

C

So if you want to use mvvm cross, if you want to use the mobile blazer bindings, if you want to use, you know maui and everything it offers, if you want to use prism or a mix of any of these, if that's possible, you know it all depends uh you want to use uno, uh that's your prerogative. I don't think microsoft's holding you uh hostage to their technology alone.

D

Yeah I mean the I think that so yeah circling back to uno, um I haven't played around with it either. It is built by developers based in montreal, but uh yeah, essentially uno. You can think of uno as similar to xamarin forms, and that is it's a cross-platform ui framework.

D

It actually runs on xamarin, just like xamarin forms does, but the difference is that uno uses uwp xaml, and so the really cool benefit with uno is. If you already have a uwp app, you could essentially leverage that same app that already exists for uvp and make it cross-platform using uno.

D

One of the big things that uno has that xamarin forms does not is, uh I guess, we'll call it web support. I'm I'm not a not a, not a web developer. I already actually just built my my first website ever.

D

I had to do an internal site, that's very, very ugly, there's no css or anything, but it's functional. I built that this week, yeah the cool thing with uno. Is it leverages uh uh wasm so that same cross-platform app that we'll be running on uwp on ios, on android, using xamarin will also run on or in the browser, and I'm not sure what engine they're using? But if it's wasm I would imagine they're doing something.

D

Are they doing stuff with blazer? um I don't.

C

Know, but did the I know that I was. I was in our meeting about blazer a few months ago which, by the way, there's a an uh indian.net consortium talk about blazer. So if you look at my other meetup, which is the indy net consortium, we uh nick george did a whole thing about blazer and I don't know if the two teams have worked together.

C

um But if you take a look, he does talk about that.

D

Right on, but uh yeah, who knows great stuff and yeah kind of, like you mentioned earlier, ari, like it leverages and uses all of the microsoft suites. So of course, microsoft is not just okay with it, but encourages it because it gets more people using the dot-net stack using the.net tooling, like visual studio, so yeah. It's definitely a welcome addition to the net community, and this is this.

C

Is cool like I'm rushing my memory about uno, I mean, if you're looking at it, and I don't know.

D

How you say your name.

C

um I don't know if objective c yeah that what that is a brutal language, all right, the uh I'm trying to remember what I was gonna say um yeah. I I think you know what what xamarin forms really offers you and it seems like uno, would offer you as well. Yeah is yet another presentation, tier. You know a solution for your presentation, so yeah.

C

I think if you have everything architected properly for your business objects and just the supporting architecture, your application, you can mess around with the the the presentation to your almost any way you see fit. You know as long as you're, not hard coding. You know like hard dependencies on everything. If you code with intent, I think you can play with all of these and see what works best for you. uh Microsoft seems to be going in the blazer direction and by the way we need to get started here in a minute.

C

But I'll answer um I just looked at the clock, I'm like oh no. I plan on starting at 6, 40.

D

Welcome to today's meetup about uno.

C

About everything yeah, you know it's it's you know we call this the 42 presentation um yeah, but again I I gotta stress that you're not stuck with blazer or xamarin forms, yeah or microsoft's technologies.

C

When it comes to that presentation tier you know, you're you're stuck with the platform in the sense that you know or with the framework, but that's not a bad thing. You wouldn't you can't call yourself yourself, a microsoft developer and not buy into the framework. You know, that's that's just part of it, but I think if you treat these as presentee to your options, then it might work out better. But that's my thought on that. So so let me get started. um Welcome to indie xamarin um happy to have everyone here.

C

uh We got a great presentation lined up for you tonight uh with brandon minnick is, am I saying it right? Brandon, yep, awesome uh and brandon knows everything about everything and he's gonna present. uh um So we meet every month on the fourth wednesday and uh we we've got a lot of topics. If you take a look at the uh the meetup site for indie xamarin, we record every one of our presentations. So if you missed anything or you're curious about what we talked about, we've covered everything not just xamarin forms, but we've covered.

C

You know why you might choose xamarin. um We've covered xamarin essentials. We've we covered.net maui last last month.

B

C

Think it was last month uh we've covered mfractor, we've covered a lot of different things, so uh you know definitely uh hit up the meetup site and you'll see the links to all that stuff. We usually post a copy of the presentation uh within a day or two of any meetup. So if your friends couldn't make it, you know no harm, no foul, you know please invite them to the to our site so that they can take a look and learn.

C

um So, uh let's see, if I can flip to my slides here, because I have all these different screens yay a special thanks to our sponsors uh sandy solutions- that's our big big, sponsor and launch fishers, which, when we're not meeting virtually, we meet at launch, which is a co-working space uh mfractor which, if you haven't used.

C

It's like resharper for, for you know, for xamarin studio for visual studio for mac. You know- or it's like resharper for for uh xamarin, really, because now mfractor's cross-platform, it's available on windows and mac.

D

And then rev debug.

C

Which, if you haven't used that it's a really cool tool, it does for debugging what resharper does for writing code and I don't mean, makes it slow. um It's a it's. A really cool tool like you could record your entire xamarin debugging session and then step through it line by line with state recorded.

C

So you can find all the exceptions that occurred and see them in a timeline just like a video player, so very cool cool stuff um tonight, uh as as usual uh we're one of the many meetups that are part of the net foundation. uh If you haven't joined the.net foundation, just definitely check this out. um They run the dotnet foundation, virtual user group, which is phenomenal.

C

uh It's there's, there's meetups, going on every single day across the globe, and the the.net foundation brings all that together in one place, so we got meetups all over the place, but you can you can sign up for the one.net foundation, virtual user group and it'll. Let you know when new topics of interest are coming out and a link to groups like ours. So you know you might be here tonight because you heard about this from the dotnet, uh virtually the net foundation virtual user group.

C

If you did come that way, yay thank you for joining. uh Please also hit us up uh on indyxamaron on meetup and uh you know subscribe to our stuff. Yeah click the bell subscribe and like kidding kidding, although I don't even I don't think you can do that when it's just a live stream like this um again very excited to have brandon here tonight.

C

If you have any questions I'll, be monitoring the the chat window, that's coming through and I'll make sure to bring those up to brandon, because he's not gonna be able to see the chat, possibly uh while he presents uh you might be able to. I think stream yard might some of it, but there's like a 20 second delay. So I can tell you once I see it, um but please feel free to answer questions or ask questions anytime during the presentation we'll try to get them answered.

C

Sometimes it might take a few minutes and with that, uh take it away. Brandon.

D

All right, perfect tabbing! I was just uh logging into the app that I'll be showing off today. So awesome. I was hoping to sneak that in really quick, so I didn't risk sharing any passwords with the world, but uh oh we're already showing my screen. Look at that whoops. Let's go back then! Okay, uh awesome thanks! So much um yeah thanks everybody for for joining us today and uh alright.

D

Thanks for for having me, it's always a pleasure hanging out with the the indie xamarin user group, and today's talk is all about architecting xamarin forms apps and actually before we get rolling uh ari. How much time do we have? When does uh when's the stream end.

C

So we usually go for an hour to an hour and a half. You know: okay, cool, that's that's about! You know what people are comfortable with. You know it is. It is getting into some people's evenings, um but you know, of course, if you miss any of it, it's all recorded so yay.

D

That's right, okay, good to know yeah just curious, because uh yeah happy to hang out with everybody is answer to answer as many questions as we can get uh like I already mentioned. If anybody has any questions drop them in the chat, uh ari you're right, I won't be able to see them because I'll have my slides up or I'll have code up so feel free to jump in uh anytime. We want to keep this as interactive as possible because yeah at the end of the day, I already I already know what I'm gonna say.

D

I already know all this stuff, so I'm more interested in uh answering your questions and trying to help connect the dots on. Maybe some new topics so feel free to drop. Any comments.

D

Aria will chime in anytime we'll have a nice conversation today, but yeah we're gonna be covering a lot of stuff, because when, when already reached out and asked me to chat about architecting xamarin forms, uh the scope of this talk in my mind, just started growing and growing, because there's so many so many buckets that architecting falls into because one of the cool things with xamarin forbes apps. Obviously is their cross-platform, they run on ios and android, and so there's you know.

D

How do you architect that to make sure we have this cross-platform experience, we have shared code and then branching out from there there's things like running ui tests. So ui tests are it's basically code, emulating a user. So if, uh if a user can tap a button, we can write a ui test that can tap a button that can enter text essentially emulate those user workflows.

D

There's unit testing, which you know we're all responsible developers. So we always write unit tests right and how do we architect our unit tests and then one of the cool things with xamarin being in c sharp and c sharp having such a strong back-end presence for things like apis and websites, web servers is we can even share code between our mobile app and our backend.

D

So there's there's gonna, be lots of layers in this stock and I'll keep an eye on the clock to make sure we can touch on everything, because I I know we chatted about maui a couple minutes ago in the intro and I included well one slide uh to give kind of a sneak peek of what architecting will look like in maui.

D

So, uh let's jump into it, uh there's a couple slides, we'll be mostly looking at code today, but what I want to point out is two links that you'll see on the bottom of just about every slide today. So the uh bottom left that is my twitter, handle the code traveler. If anybody has any questions, uh feel free to tweet at me or shoot me a direct message, my direct messages are always open. I'm always happy to chat about xamarin.net c sharp. uh I actually I used to work at xamarin.

D

That's how I made it to microsoft when they acquired us a couple years ago, and so I'm definitely a big big fan of xamarin uh and on the bottom right. This is where you can find the source code for everything we're doing today. There's a an app that I'll do a shameless plug right now called get trends that I built.

D

um Oh no, huh I hit my api limit. That's okay, because git trends has a demo mode sure.

D

Well, ironically, this is in the app store today. So I'll tell you what let's go to I'll! Just show you what fit! Trends.Com spell it right. So this was an app that I created actually about a year ago as just a an app to help me because I I love the open source community. I have tons of open source repos. I think at last count I had over 300.

D

No, it's not that not that they're all being actively maintained, but uh a lot of them are, and so one of the problems I was having with all my open source repos was just keeping up with them and how to manage them, because what I found was.

D

I had some repos that I hadn't touched in years that I found were getting a lot of traffic, so they were still getting 12 20 views a week, and this is source code that folks were referencing for their apps, and so I wanted to be able to stay on top of which open source repos of mine are getting the most traffic. That way, I can make sure to keep those up to date.

D

um So I created this app called get trends it's available in the ios and android app stores shameless plug, like I said, but um what we're going to do today is check out how I architected get trends and with everything we're chatting about today. There's always multiple solutions to any problem, uh no matter what anybody says I mean. I know uh some folks are very passionate about architecture and will subscribe to only one way of doing things like you have to do dependency injection.

D

Well, do you have to it's recommended, but no you don't have to so. What we'll do we'll walk through get trends because git trends has it's all built in xamarin forms, and it also has a suite of ui tests, a suite of unit tests and even has a cloud back end, all of it's written in c sharp, it's all sharing code amongst everything and so we'll be leveraging a lot of this code.

D

So all that to say at the bottom of the slide boop, you will see the link to the source code gettrends.com, just redirect you to it's github repo, where you can download the source code play around with it yourself. So all that to say, don't worry about memorizing anything today, because it's all open source, it's all available and everything we reference is actually linked down below. So there's some cool libraries we'll be chatting about like refit, sqlite.net poly and all of these are linked in this readme.

D

So you don't have to worry about memorizing anything taking screenshots of links, if you just remember gettrends.com. This is where we can come to find everything.

D

So with that being said, let's jump into the first architecture topic, and this is what I'm calling the mobile app architecture and, if you've, if you've done any xamarin forms development.

D

This will look very familiar to you because, like we've mentioned earlier, xamarin.forms is a cross-platform ui that runs natively on ios and android, and the way xamarin forms works to be able to still create the native ui button on ios or the android widget button on android is that in xamarin forms when we just say button, but when we compile for ios our code that we called button in xamarin.forms gets turned into the native ui button and likewise on android, that button in our uh shared code gets turned into the native android widget button.

D

um Actually it's probably been updated, android x, but anyways. um This allows us to essentially write our code once and write it on both platforms, while still getting the native performance and native look and feel, and so how do we do this? Well, we have a we'll say a common project.

D

I didn't want to say shared projects that actually means something, but a net project that is common for both ios and android, and both these ios and android we'll say head projects reference that code, so you'll see a lot of these diagrams in the slides today, where we're talking about references and literally what that means is, if we jump into our code- and we look at say ios if we click on, add references.

D

Well, there's a check box next to git trends, because the ios app is referencing the code and essentially that's how we're we're sharing our code between these two projects. Now, yes, this one's pretty basic, wanted to share just in case anybody's new to xamarin or xamarin forms, uh but I also wanted to talk about my recommendations for this, which is to use a dotnet standard, 2.1 library, uh there's a couple rules of well, I guess first, what is net standard right?

D

uh Dot net standard is a a way to make a c-sharp project that has no run time, so why would we ever want to do that? Well, in this case, we want to leverage the native runtimes. No thanks. Dropbox we want to leverage the native runtimes like for ios. We want to use the mono's ios runtime and likewise on android. We want to use its mono android runtime, and so what dot standard gives us?

D

The ability to do is create a project that has no runtime included and really all it is, is just a giant interface that says here are all these apis. You can access specifically.net api, so things like system.datetime, um well, any system, library really all those ones. We know and love and there's a contract in place that I xamarin ios, xamarin, android.net, core, 3, etc, have all said.

D

We will support all the available interfaces in or all the available apis in this specific.net standard. So um so yeah basically.net standard is just a project that can't do anything, but we can run it on different platforms like xamarin.ios same and android.net, core, etc, and so when we are creating our standard performance projects, my recommendation to you is to create a dot-net standard library and then for the version of it crank it all the way up.

D

The the latest greatest version is done at standard 2.1, which I'm referencing here, and that's really all you need to do to get started because down at standard 2.1, the bigger what the number means, the bigger the number, the more apis are available. So down at standard 2.1 has the latest greatest stuff, um like iasync, innumerable uh async streams, all that fun stuff and again, when that code gets compiled down, it'll run using the native runtimes for xamarin, ios name and android.

D

Now you can also use what's called a shared project. The difference between a.net standard project and a shared project is the shared project.

D

Let me rephrase that the difference is that the net standard project, when you compile it it'll, generate a dll. So, for example, if we jump into our code- and let's see if I dig into the outputs for my.net standard library there, it is, we have gittrends.dll. So this is a.net. Standard project creates a dll again, it doesn't have any runtime.

D

This dll just gets sucked up by whatever runtime we're using in this case uh xamarin android, here boop, and so uh with the difference with there we go with a shared project, as a shared project does not generate a dll. The shared project actually is subsumed from whatever project is referencing it.

D

So what we would see if, if this get trends library was a shared project, it would not generate a dll and when we build for ios all the code that exists in this shared project would just be built as if it existed in this git trends.ios library, and so there's some pros and cons there. But the recommend. The reason I recommend using.net standard is dotnet standard, allows you to install nuget packages directly into the library.

D

So if we look at our code here, our.net standard, I've got all these awesome nuget packages from amazing, uh open source libraries. Here. If this was a shared project, we would see zero nuget packages because it's impossible to add them to a shared project and the tough thing there- and this is the reason why I recommend using down at standard- is if we can't add dot.

D

If we can't add nuget packages to our shared code well, then we have to make sure both our ios projects, which we can add new, get packages to and our android projects. We can also add new packages to are in lockstep, and I've been burned with shared projects. A couple times where I updated xamarin forms nuget package in the ios project and forgot to do it in android, went to run the same app that worked beautifully on ios on android and all of a sudden.

D

Nothing worked so uh you'll hit weird little things like that, so you can use them dot or shared projects are super powerful, but it's one of those things with great power comes great responsibility.

D

So if you don't, if you're new to net, stick to dot-net standard for now, if you have you shared projects before, if you understand those uh the power which that power really just means more opportunities for you to shoot yourself in the foot by accident, which I definitely still uh do occasionally, if you understand all those things, go for it with a shared project, but if you're new to xamarin.net it's hard to make similar mistakes like that with dot-net standards, so yeah so right.

D

Well, all of our xamarin forms code lives in this.net, standard library and, if you're, only targeting ios and android crank that up all the way to 2.1.

D

Now the caveat is what, if we're also targeting uwp well, uwp doesn't yet support net standard 2.1. So same same for everything we talked about, I still recommend using standard over a shared project, but in this case our peak.net standard level will be 2.0, which is still good, still has everything we get and we'll say everything.

D

We need to create a great app it's just about a year behind and it doesn't have cool things like async streams or iasync, innumerable which I know I fell in love with uh when they debuted so just something to keep in mind. If we are targeting uwp for now, we can only go up to dotnet center 2.0, but if we're only doing ios and android crank it all the way up to dotnet standard 2.1 all right, how are we doing on um questions ari anything coming in any anybody? Any fans of shared projects out there.

C

No, you know I, I have something to add about shared projects. I used to always use shared projects because going the pcl route, which is what you used to have to do. Yes and shared projects got you around that, um but that's where I learned about exactly what you said, which is uh the the nougat versioning nightmare. You know like hey, you know I referenced one version of uh let's say: newton's off json example uh in ios and use the feature it has, but I use the wrong an older version.

C

It didn't have that feature in android. Well, visual studio doesn't know what doesn't exist in each project, so you know the the shared project overlays those projects. So basically it's like it all. The files that are in that that shared project are part of both the projects that are referencing it, so it's kind of like they're injected as if they were part of that project. The problem is you run into a lot of compatibility issues there you also run into- and I 100 agree about going the dot-net standard route.

C

uh Well, we're getting some questions in yeah um all right. The uh a couple things there. One I think you'd be safe, going.net, standard, 2.1 and ios and android, but you got to have the shared libraries have to be the lowest common denominator, so you'd have to have.net standard 2.0 for that head project that you're talking about that contains everything.

C

um But you know the ios and android, I think could be 2.1 because they can reference 2.0. But you can't go the other way around, um but be very careful if you go the shared project route you're going to have to get used to compiler directives, because we're going to have to know about if the platform is ios run this code. If the platform's android run this code and your code gets real ugly real fast.

C

So if you can stay away from shared projects, I I've learned to stay away from them questions um when asp nerd asks. When will we move from standard to core.

D

Ooh, that's a good question and we'll touch on that later. When we talk about uh donette6 and donette maui towards the end.

C

Yeah and at least on vs4 max so visual studio for mac, so john asks at least on visual studio for mac. I don't think there's even a template for a shared project anymore. I imagine you can still create one, but it might not be so straightforward.

C

Well, my thought there is good uh explaining shared projects, it's magic and I think they had their place when you had pcl, but I don't think they had their place now. That.Net standard exists.

D

Well funny you mentioned that, because later I'll actually give you a recommendation when I do recommend shared projects: okay, okay, we'll just leave that as a little teaser that's coming coming shortly and we got one more.

C

Does get trends use any caching like in memory? Caching, um for example, when internet connectivity is not good, so I imagine yeah. Well, you know offline good offline patterns, yeah.

C

Of course, um let's see, I'm gonna, I'm gonna post um a good chat or a good article uh about resilient network connectivity in xamarin I'll put that in the in the comments for everyone, uh because you learn about refit and poly and acavatch how those things can really help you and if matt su stucup was here, he talked probably about the uh the built-in stuff in azure that handles online offline. Sync, for you.

D

Yeah and actually we'll be touching on poly and refit today, as well, all all the good things all right, all these little teasers, it's like we planned this.

C

Any other questions we have no other questions right now, so yay all right.

D

Yeah, so just uh that question about caching yep there's: this is using a sqlite database which we'll also touch on in just a minute that um stores the data every time we pull it down from github. So this this app just pulls all the github apis. uh The funny thing here is I'll. Show you the secret behind it is.

D

This is just using github apis and all the github apis are available to anybody, and so what I did I'm not signed in dang it well, if I was signed in, we would see a option here to look at. um I think it's it's called insights and on insights is where you can see these numbers so like how we'll get meta with this. We'll look at the data from my get trends repo, but all these daily views daily clones unique views, unique loans.

D

All this data is just coming from the github apis and it's available on the website, but it's buried in this insights tab and the reason I want to create this or reason I did create this app was. I didn't want to have to click on 100 repos every week or so dig into their insights, tabs and figure it out, but uh yeah. So, let's, let's power forward, we got tons of fun stuff to talk about today.

D

So that's just kind of the overview of how you want to architect your app, um but now I want to focus in on specifically in on xamarin forms. So I think the first thing that is important to talk about is mvvm.

D

So if you've never heard of mvvm, it stands for model view view model and, like any architecture, you know maybe you've heard of mvc model view controller. Mv. Excuse me, mvu model view update. It's basically just a way to architect your app that separates concerns so model in mvvm. Let's get that oh model.

D

This is essentially our pocos, our plain old c-sharp objects. This is just uh structure of our data and I'll show you what that looks like in code in just a minute, but basically models. Don't do anything, there's no logic in them.

D

But, for example, when we're pulling down data from the github api github api is going to give me back data for like the total number of views, and that has a model which is a list of int, probably a list of it, uh but that that model just defines the data the shape of the data. If you will, it doesn't actually do anything, um the view is just the ui. So again the ui doesn't do anything. It just knows like in this case, to draw this rectangle color it uh color the background white.

D

Add this image add some text, but the view doesn't even know what image is going to appear here. The view doesn't know what numbers or what words for the text are going to appear. It just knows going to give me an image, url and somebody's going to give me some text and when they do, this is where I put them. So our view um is intentionally.

D

uh I don't want to say dumb, but we'll say out of the loop when it comes to um things like uh what happens when I do it, when the user does pull to refresh or anything like that, it's literally just drawing things onto a screen and then the view model in mvvm.

D

This is where all that logic takes place, so the model defines the shape of our data. The view describes how our ui looks and the view model is kind of that glue, holding everything together, because the view model knows how to take action. So, for example, when I do that pull to refresh it's the view model that knows, okay, I've got to go hit the github apis and here's how I surface the data to the view so that the view can drop on the screen now.

D

The reason we recommend mvvm is well partly because xamarin forms was literally architected around the mvvm paradigm. So it's literally the de facto architecture I mean you can you can have libraries like uh the reactive extensions will allow you to do things that are like closer to mvu. uh Some people will claim it's not pure mvu, but uh there are ways to do other architectures, but the intent, the creation of xamarin forms, is all around the assumption that we'd be using the mvvm architecture.

D

So let me show you what that looks like in the code and so in my code, let's see if we can make that a little bit bigger there.

D

So, for example, what does a model look like? Well, I have a model here. This is my referring sites model. Let's make this a little bit bigger and just uh wondering what a referring site is. uh The github apis will also tell you where that traffic's coming from so in this case, this github repo is getting some traffic from other github uh repositories that are probably pointing to this repo. uh Somebody did a blog post that points to this repo or just a plain old google search that landed somebody in this repo.

D

So that's a referring site and let's see this is this- is a little complicated. Let's jump down to the base class here, ah much better, so this model just has a couple properties in it. So the date time for when the data was downloaded, the name of the referrer, which would be google or github, uh we make sure that the referrer's uri is valid. That's what the code in this constructor is doing just to make sure um well. This was something unique to the way. Github gives me the data.

D

Sometimes it includes the full url, but for popular sites like google, it just says google, so I have to do some logic like adding.com if it's missing and making sure everything's, https and then and also the it gives me the uri. So if, if we're back looking at the referring sites in this app, if I tap on, I write code, sometimes it'll, take me to the uri and and that's it but yeah.

D

I tapped on that thing or I tapped on that uh that item in the list and we navigated to its url, but none of that logic happens in the view. This view is literally just a couple properties that.

D

Declare the shape of the data, so every every referring site will have these specific properties in it and then it's the up to the view model on how it handles that.

D

Now, if we look at a view model, let's actually check out, let's hide that don't want to give away what that common. Folder is just yet if we look at a view model, for example, that referring sites view model. So this is the this view. Model is the business logic behind this page. So things like what happens when I do pull to refresh well, there's somewhere in here a refresh command whoops!

D

Oh no! There we go!

D

That's what I wanted, so this command executes anytime. We do a pull refresh. So if I were to put a breakpoint right here, do a pull to refresh the debugger hits this breakpoint. So it's the view model that knows what to do anytime. The user takes an action and there's more code in here about you know what what do we do if the user clicks on it?

D

It timed out, because I had hit the breakpoint and that's fine, but the view model is what handles all the logic. So it gets things like this referring site list, so that was the model we were just looking at again. The model doesn't do anything, but it just defines the shape of our data and it's the view model that handles and knows what to do with that data and what was the other thing? The view of course model view view model.

D

So the view this might look a little funny to folks who have been doing xamarin for a while. I actually I create this pages folder in my app. um I started this way back in the day, because every let's do referring sites page, because everything in xamarin is a it's a content page. So it's like oh well I'll, make a folder called pages and then later it's like, oh well. Technically, this is the view in mvvm, so you'll see probably most projects we'll call this folder views.

D

I actually have a views folder down here, which makes it even more confusing, but these views are basically views for a specific page. So the referring sites has some specific views for it, but if we look at this um well, first of all, I do all my ui in c-sharp, which is has come a long way. We recently got c sharp, markup extensions and xamarin forms 4.6, which are basically extension methods that allow us to create uh that give us a fluent syntax for our ui.

D

So, for example, I can create a collection view and I can say, dot bind, and this should look and feel familiar if you've done any work in xaml, because yeah we're just creating bindings uh we're putting things in a grid. One of the things I love about this c sharp markup is. It allows me to define my rows by literally giving them a name uh if, if we're working in zamble, you just have to say row 0 is auto and row.

D

2 has an absolute length of well in this case, whatever the shadow height is and bro, I guess no. This is row two there we go see. That's why I don't like or that's why I do like the fact that I can define them, because now I can define each row and I don't have to remember- is this row zero? I mean it's the first row, but it's row zero because counting starts at zero in c sharp and I would always get that confused in my head and anytime, I would create a new ui.

D

There was always something in the wrong quadrant because of just trying to remember all that stuff, so yeah with uh these markup extensions, I can just say this row is called title. uh This row is called title shadow. This row is called list and then, when I put things into the grid, for example like this reviewing referring sites refresh view, I can just tell it it's going into the title shadow row: it's going to span three columns or span three rows and then span all the columns which is really cool.

D

But when we look at this, there's no logic in here. This is just drawing onto the screen. So the view just knows how to draw onto the screen and all these bindings you'll notice are pointing back to that view model.

D

So things like this, this refresh view whenever it's commanded to refresh it actually shouts out, to my view, model specifically to this refresh command that exists in the view model, and so this is where we were a minute ago with the refresh command, where we set that breakpoint down here so yeah what ends up happening with mvvm.

D

We have these bindings that are essentially tie our view model and our views together. So the view doesn't know what to do it. All it knows to do is any time uh portal refresh happens from the user. It just shouts out to the viewmodel and says: hey, hey, refresh command, execute, but that's all it knows. It doesn't know anything more than there's something called refresh command and it's the view model that handles all that logic, so essentially mvvm.

D

What is it it's? We have a dumb ui that we call the view that doesn't know how to do anything. It just knows how to shout to its associated view model. The view model handles all the logic and the view model can also shout back for when things are updated. So when those integer values for the total view counts, get updated, the view model can shout back and say, hey view.

D

I've updated these and they're always listening to each other and when the view hears that it goes. Oh thanks for that data. I know to I know where to draw this on the screen, and so what this allows us to do and the reason we recommend it, because you might be looking at this and going gosh. That is a lot of overhead and you're not wrong, but the benefit here is.

D

Let's say I wanted to redo this entire ui and ironically, this was an effort that we did a good friend of mine, luis pujos, reached out to me after I initially pushed this app to the app store, because, if I'm being honest, the first version that I released the app store looked terrible, uh I'm not a designer uh everything. Functionally was great. There was no bugs no crashes, so hey. I take that as a win, but didn't look great uh and my buddy luis who's. Also a xamarin developer in the xamarin community he reached out.

D

I was like hey. uh Can I help you make this look like a real app should and he was the one that came up with these this layout and uh what that meant was. I basically selected all the code in my referring sites page and said, delete and then using luis's templates.

D

We recreated it and what's great about mvvm is I could do that and I didn't have to touch any code in my view model, so I didn't have to change anything in the logic I just got to redraw or tell my app how to draw this new ui, and so that's the real benefit of mvvm. Is we have these separations of concerns?

D

So if we want to rip out a layer in our app we can if we want to change the view, if we want to do a version, 2.0 update like we did with git trends, we can do that and we don't have to touch any of the logic and we know as long as we wire up those bindings in the new ui. It's just gonna work, because all that view model logic is still there and we hadn't. We didn't touch it.

D

There's no risk at accidentally deleting logic code, there's no risk at accidentally, adding bugs into our view model, because it had no idea that we changed the ui and that's and that's the point of mpvm. So mbvm gives us that flexibility. It gives us a separation of concerns and I'll say, most importantly to me. It also helps us define where files live and what's nice, with xamarin forms, essentially going with mvvm as a de facto architecture is that I can open just about any xamarin forms.

D

App, there's tons of xamarin forms app open sourced on github and when I pop them open- and I want to see something like you know- maybe there's an app that already made. That uh is doing something really cool like he created this really cool, looking ui. Well, I know when I open up aria's app and look at the code.

D

I just look for a folder called views or, if you're weird like me and name it pages, I can look for these folders and I can figure out pretty quickly where that code lives that already made and figure out what he's doing and likewise, if maybe the app is also doing something really cool like it's using um a azure api that I've never tried out like I already was talking about online offline sync and all that cool stuff. Well, how is he doing that?

D

Well, I know that's could be in the view models, or maybe it's in the service layer which we'll talk about in a second okay. So that's yeah.

C

To add there uh and thanks this is really interesting. um The I see that you're talking about how you're building your ui via code instead of xaml. So could you speak a bit to the pros and cons of that? You know if there's a performance impact and so forth. Oh good question.

D

Well, I think some context is important here. I I never did any. I was never a web developer and I never did any winforms or wpf before I started doing xamarin, and so when I started working at xamarin. Actually, the reason why I even applied for a job at xamarin was, I knew c sharp. I I wanted to make ios and android apps, but I didn't want to learn objective-c and I didn't want to have to spend the time to rewrite the app again in java.

D

So what I heard about this weird x-zamarin company that was on the west coast, I was like oh, oh wow, yeah, that's really cool and the apps will totally native great. So when I joined xamarin, xero forms wasn't really a thing yet, and we did everything in c sharp because well, that's kind of the value prop of xamarin is we could do everything in c sharp.

D

We never have to touch objective c, swift, java, kotlin, etc, and so when xamarin forms came out, actually it was initially all c-sharp, very initially I should say, and so when I started learning xamarin forms. I just start. I just kept using c sharp and it ended up being.

D

Xaml has since become the more popular way to create uis, I think probably 75 percent of xamarin forms developers use xaml. So if you do too you're not alone, but for me I had never used ammo before and I was struggling enough just to learn mobile at the time I was still new to ios and android. There's 100 000 apis that are unique to both platforms and you have to learn the mobile life cycle and for me I didn't want to add xaml to that learning curve.

D

And so um that's why I use c sharp and circling back to your question. There is a we'll say, a slight performance benefit for doing things in c, sharp, because c sharp's, all ahead of time compiled, whereas xaml is just in time compiled. So what that means is when I launch a new page. So, for example, I click this referring sites button boop that code to display this page to run this and the associated business logic everything's in c. So it's all been compiled ahead of time.

D

uh If this was in xaml, the xaml parser has to first essentially transpile that xaml code into c sharp, and then it can generate it. So you get a little performance bump by doing everything in c sharp. Now that being said, a couple years ago we came out with the xaml compiler or xaml c yeah and xample c helps a lot now totally anecdotal. When zamblc first came out.

D

I was like how good is this right, so I I made a kind of simple dummy app and did that that test, just like this, like tap a button that launches a new page and without xaml c it took about 300 milliseconds for that page to load, when I enabled xaml c that dropped in half so about 150 milliseconds. So that's totally anecdotal.

D

That is gonna that time difference is gonna, be different for everybody, because everybody's sample pages are different and so more complicated uis are gonna, take longer to load, no matter what you're doing and or whether you're doing c sharp example, but xaml c still doesn't compile all the xaml. uh It compiles a lot of it, but it doesn't cover everything. So c sharp will always have that performance benefit over xaml.

D

Now all that being said, if your app's running fine in xaml, don't don't turn around and change it. I mean if there is a weird case where you're seeing performance problems and you think you can squeak out some more performance by uh porting it to c-sharp from xaml it's worth a shot, but uh the the difference is, I won't say negligible, but I mean we're talking in the milliseconds uh or in the tens of milliseconds between c sharp and sample. So that is um one benefit.

D

uh The other things I actually let me show you because I did a a talk recently on all this fun stuff. Let's see where I highlighted some of them here so sample first c sharp. So that's the performance we were talking about. uh Refactoring and maintenance is nice, because I can do things like this and see sharpware.

D

If I'm in my viewmodel- and I want to change the name of this command, I can just say rename- um maybe call it new refresh command and then, if I flip back to the sites page, because this is all type safe, it's all in c sharp. It's already been updated, whereas if I did that and my ui was in xaml, then the xaml doesn't automatically get updated because it's xaml is essentially just a hard coded string. So we have to remember to update all our bindings if we change any of the property names.

D

uh Likewise, if you're ever jumping into a new code base, it's really nice when everything's in c sharp, because I can say, find references and I can find find out exactly where this empty data view. Title is being used in both my view and also on that page. So it makes it really easy to come up to speed where, if everything's in xaml, then the best way I found and if anybody knows a better way to search xaml or to search symbol bindings.

D

I should say: let me know, but the best way I figured out how to do. It is just to literally use the search bar here. So we get a little bit refactoring performance benefits. uh Then there's some silly things like. I never understood this syntax, like this uh automation id just to use a constant so like in c sharp, it's a string constant. So I can just say: class name, dot, string, constant name, whereas in xaml it's x, colon static and then I have to include the namespace, but it's not even namespace.classname.

D

It's namespace colon classname- and I I don't do xaml enough to remember this, and so I hate this because I always have to google it every time like. How do I make a constant in sample right then yeah once you.

C

A

C

Right answer, your question: um I know resharper and impractor will take care of. Like the refactoring point does work. uh I haven't tried it without resharper because, like I always have resharper installed, and I already have impractor so yeah. Maybe I'm spoiled you know, but it will handle those bindings for you and it'll even give you the uh the intellisense.

D

To let you choose that stuff yeah, that's a good point yeah! If anybody's listening and you make your uis and sample, make sure you're. Also using empractor, I found empractors, essentially a a must have, and yes, it costs money, but it's one of those productivity things. That's just worth it and plus you can probably get your company to pay for it too. But um the last thing here and then we'll get back on the rails to architecture.

D

Is these converters, so creating converters and samples is really annoying because you have to create a whole separate class. uh This I value converter is not type safe, so everything's an object, object, value, object, parameter and then you have to cast it.

D

Hopefully you got the casting right, otherwise, your app's going to crash at runtime, whereas in c sharp in our bindings we can just say, convert and then just pass in a function right here, that's already type saved, because I told the button that this is or I told the binding that this is a button to and then int and bool are the two, uh the source and the destination.

D

So I have type safe bindings that I can just do in line without having to create a whole separate class, which is really nice.

C

And john uh bachelor points out that jetbrains writer does a great job with xaml bindings, uh uh of course, jetbrains resharper same company, so uh ryder is very cool and you get a lot of the resharper capabilities built into the ide there. You go zama fans.

C

Writer is not free right. I don't think it is.

D

I don't know I I hear great things I honestly haven't tried it really good yeah. I mean my thing is mostly just I work at microsoft, and so I try to stay on the the preview, slash dog, fooding channel, of visual studio over studio for mac, because I want to try and catch all the bugs before or help catch bugs. I should say, because I personally won't catch all of them, but before it gets to stable.

D

So that's one of my say altruistic things that I try to do to help out the community, but all good questions.

C

Yeah ryder's 139, okay, so there you go.

D

All right, uh so the last thing I want to touch on in xamarin forms and really just in mvvm in general, is I do recommend dependency injection. Now I mentioned at the top of the show that some people are uh almost religious and they're fervor for for slash against uh dependency injection.

D

I personally, I used to be not against dependency injection, but service location uh was so easy. I was like this is fine and it wasn't until I got into unit testing, which we'll touch on a little later, that I really fell in love with dependency injection and, if you've never heard of dependency injection or maybe you've, never done it in the xamarin forms app.

D

It's there's actually a nuget package made by microsoft.

D

That, uh oh, hey, I'm not even using that here, I'm using autofaq, so microsoft also makes a dependency injection library the one I'm using in this is auto fact, but basically what you do uh if we look at my view, model you'll notice, this constructor has tons of things being injected into it, and these are all different services.

D

uh There's also, uh this is xamarin essentials interfaces which I'll talk about a little bit later, but what dependency ejection does is, instead of using static classes, we can inject that into the constructor, so any outside class that we're going to be referencing. We can pass in and the reason this is or one of the reasons this is nice is we can kind of set it and forget it. So I have this dependency injection container, where I just register all these types.

D

So if any class wants to use app info implementation um or if they want to use and my analytics service or background fetch service they're all registered in this container, and what that allows me to do- is where's a good example for um grabbing that down here after the splash screen.

D

Paint page that's right page there it is so to initialize my repository page, I can just say container.resolve and then it initializes it for me. So what's actually going on here with dependency injection, is this repository page has all these dependencies, but everything is registered ahead of time. So when I say resolve all these implementations are just added in for me, so I don't have to worry about newing up a main thread class or first run service or any of this stuff.

D

It's all just done for me, and a big benefit is that we can do things like single licenses for some of my services, so anyways get a little bit ahead, but uh dependency ejection is something that I've recently fell in love with. I recommend it. It is complicated, though so, if you're a new developer, uh if you're new to csharp.new.net, it's probably easier to stick with service location or the service locator for getting started.

D

But for me once I started being better about unit testing my zero force projects then depending actually became essentially a must-have tool.

C

I'd love to add something to that uh one thing: service, location and dependency injection to do is thinking abstractly and coding with intent, so you'll get out of. It is building the interface first figure out. What's the intent of what your code should be before you think about implementation, and then you'll actually write your implementation. So, as you come up with better implementations, at least your intent stays the same because you know if you're doing crud, you know your c r. U and d really aren't going to change, but your implementation might so.

C

Your interface will have your create, read, update and delete uh kind of yeah just but then.

B

C

Be able to swap that out, as you get better or as you find issues so, whether you use service location uh or use dependency, injection ioc uh you'll be in good shape, just architecturally by starting with that pattern.

D

Yeah and like I said it is, it is tough. You know it took me a couple years of doing c sharp before I felt comfortable with this dependency magic or dependency injection magic, but yeah once once you can kind of wrap your head around it. It is beautiful because yeah there's even things where, like, um if a class references, a class, it references, another class versus another class that references back to the same class.

D

Well, that's called a circular dependence and if you try to do that with dependency injection, that dependency injection container will literally blow up be like whoa whoa.

D

Nope can't do this and, like I already was saying it, it forces you to essentially rethink how you architect and write your codes to avoid those circular dependencies and really what that comes down to why those are bad is those can lead to memory leaks and worse, oh, maybe it's worse depending on your take on it, but it could also lead to what we call spaghetti code where um the code just becomes unmaintainable so yeah forcing yourself to think about that dependency, injection architecture, essentially yeah.

D

It helps you write cleaner code well said yep all right, so something else we have in our xamarin forms architecture is what we call the service layer, so we mentioned earlier the view model that does all the work like when I click a button. It's the view model that knows what to do after the click or when I do a pull refresh. It's the view model that knows how to execute that refresh.

D

Well, if there's code, that I say is agnostic to a view model where, for example, writing code that hits the github apis well, I have a couple view models that all need to hit the github apis. So that's not really specific to a view model. Then we can put that in what's called a service layer and the benefit there is well there's a couple benefits the one one big one being that separation of concerns, so uh that single responsibility principle where each class is responsible for essentially one thing.

D

So I have a class that's responsible for interacting with the github apis, and I can now share that class between multiple view models. I don't have to rewrite any code and so to kind of simplify this a little bit one of the ways I think of like when I think of, should this go in the service layer. I ask myself: uh will I be using this? In other view, models well multiple view models need to use this code. Then I break it out and I put it in the service layer.

D

And so what that looks like in git trends is I have a folder called services. So again, these are all things that are agnostic to specific pages so, like um I use app center to do uh to track uh crashes and get crash reports and uh timing things I use like we mentioned the github apis, and these are all this is all code that accesses the github api.

D

So it's very easy to know which or find those services, because if I want to access github apis, I can scan through my services folder and then here's the v3 apis, the graphql apis for github, and I can quickly figure it out what I need to reference and if I didn't have them in the services layer, then I would essentially be duplicating this code across multiple view models, and so that's always my trigger is, if I find myself duplicating code between view models break it out, put it into my service layer and something I try to do well, I do do is all of my services, they're all we'll say single instances, because that's what it says here, but basically for all types of purposes.

D

These are singletons and the dependency injection container takes care of all this. For me, which is great um but what's nice, is when I have two view models that need to use the github api v3 service.

D

Well, I don't need two separate instances of these, because my services don't maintain any state, there's nothing in the service that is specific to things like ah the app went away, uh but is the view refreshing is the view busy, all that's handled by the view model. These are all stateless that just know how to hit an api and get data back or do some image. Caching or translate languages and skit trends handles a couple different languages like spanish and dutch and portuguese shout out to the community. They added all these awesome languages.

D

For me, since the app's open source, I just got all these awesome. Full requests that were like hey, I speak spanish, wouldn't it be cool if we had spanish language support, I was like yeah. That's awesome. I don't do you mind, opening a poll request, so we are still looking for chinese. If you know anybody that speaks chinese, send it my way, because I have a lot of chinese users for git trends, but we don't have a chinese translation yet so anyways yeah services, they are potentially view model agnostic.

D

They can be used by multiple view models. I recommend keeping them stateless, so they can stay as singleton, so we don't have to create a new instance. Every time we want to say pull something from a github api, but one of the things we mentioned at the top of the show a little teaser is poly and refit. These are two must-have libraries. um If you've never heard of them, poly allows us to do automatic, retries and exception handling.

D

So when you think about resiliency like we were talking about earlier, if let's say well before kovid, let's say I'm on the bus and the bus is switching cell phone towers or it's going through tunnels. Well, the user might lose that internet connection briefly, which would cause that api request to fail well with poly, I can tell it to automatically retry, and so, when that happens, my code automatically hits the api again and you can set how many retries you want.

D

I think my default's three you can set things like um how long to wait between retries, because you don't want to retry right away because computers operate in the nanosecond scale, the millisecond scale. So if I couldn't connect to the cell phone tower uh a millisecond ago, there's a likely chance that I won't be able to connect to it again right now and so just to show you what that looks like in the code.

C

And we do have a question: oh yeah shoot um is that so mark is asking. Is the view model responsible for getting and or storing the data that is in the model and shown in the view, or that the model is really fully.

D

uh You were exactly right in the first part, the view model is responsible for getting the data and then um it essentially it jams that data into a model and the view knows what to expect from that model. So they all kind of work seamlessly together but yeah. It's all up to the view model to do the heavy lifting the model's just kind of hanging out and going hey here's what the data needs to look like. Let me know when you got it and when you're ready for me.

D

So here's a example and code of what poly looks like um this is a method I wrote called attempted retry, where kind of like it sounds: it'll, wait and retry automatically. So I use this for all. My api calls like if I jump back to.

D

So like this api service, we'll always call attempt and retry when I want to get. For example, the repository view statistics and so what's nice is, I tell poly the number of retries I set my default to three and because we don't want it to retry right away because again, if we couldn't connect to the internet a millisecond ago, probably won't be able to connect to it now a millisecond later and what I do is what's called an exponential back off.

D

So I tell it to it passes in the number of retries, but it'll weight, two to the power of the number of retries, so this code is basically just doing to the power of n. So this means the first retry will wait two seconds between when that api call failed to when it tries again, because that's two to the one, uh the second time it's two to the two squared. So that's four, so it'll wait four seconds and then the last time will be eight seconds in between and so hopefully eight seconds.

D

If eight seconds have gone by, the user is reconnected to a cell phone tower and if not, then we know for sure that their internet connection is dead.

D

The other nice thing with poly is, we can tell it which exceptions to handle and which ones not to so I created I created this little local function that checks to see if this api api exception is forbidden or unauthorized, because if I try to connect to the github apis- and maybe my tokens expired well- I'm not going to retry that because the tokens expired so.

C

D

Point in retrying, because I already know that this has failed because of well, not necessarily because of the user, but because of something on the user side or the client side. So yeah I just checked to see. If, if that is, if it was forbidden, if it was unauthorized, then I don't retry and then all the other ones yeah retry.

D

So if there was a timeout exception or something like that and what's nice is, I now have this static method that I can just leverage for all my apis like this, where I just say, attempt and retry and poly will handle all of that for me, so huge fan of poly and the other one I recommended here is refit refit.

D

Oh my gosh, I don't. I laugh when I think about life before I found refit because refit, if you have to access apis, see there we go so refit. Lets us define our apis the same way if you've ever done anything in asp.net core. If you created an api or controller in asp.net, core or asp. framework, you you're familiar with this attribute syntax, and so all I have to do is tell refit to say.

D

If I want to get the repository view, statistics well, here's the api I want to hit I've already told it the base url by the way. So it knows the github api, slash, repos owner and actually you'll notice, owner repo are in brackets.

D

Well, that's because I need the user to pass those in, and so this is it this one line of code. I can call from my app just like this and it knows to make a git request that github api. So before this I had a a base class that was kind of generic and static, and I would just copy paste that everywhere for doing git apis in c sharp.

D

Once I found refit, I just deleted that because I can just easily define my api endpoints like this, so huge huge fan of refit here you can pass in headers. You can do things like um you can even pass in headers at runtime all sorts of cool tricks. You can do with refit and it'll save you tons and tons of time trying to manually write that code.

D

Oh all, right, hey all right! I noticed we have about eight minutes before the top of the hour. um Is it cool if we keep going a little longer.

C

uh I think that it's okay, you know, I mean we're. Recording the whole thing, heads up, yeah, the refit stuff um and the oh. We got a question come in. uh Does it? Oh, that's a good question. Asp.Net. I heard nerd.

C

Sorry asp.net. Yes, I'm gonna call you ask nerd, I don't know how to you know. Maybe his name isn't asp.net. Maybe it's not asp nerd. I keep want to say asp.net dang it. What's the question. He said. How does reef handle connections? Does it make him reuse? One client per base, url ooh.

D

Yeah, you nailed it uh if I zoom out, so I have this github api client, which is defined up here. So this is that interface that refit interface and yeah every action that's taken using this client will reuse the same http client with the same base. Url and let's see where do I define this?

D

C

A

C

D

So I have some code here that uh creates http client, but that client gets passed in um you just say restservice.4 that that's the refit api we're just looking at and then you pass in your http client and that's where I pass in my github api url but yeah. No, you nailed it. We give it what base url and it'll reuse http client, because if anybody's listening and you are disposing of http client, don't hdecline is meant to be reused, never dispose of it. um There's whole things with sockets, but we'll say trust me, reuse.

D

Your http client.

C

And that's because it's a re-entrant object, that's the that's the key I'll post, the uh I'll post, that link http client as re-entrant um yeah, if you're using a using around it you've been using it wrong, which I didn't know until a couple years ago, when I read about it, um I also put uh the links to the uh resilient network handling and or resilient network connections in xamarin.

C

Those are on the indie xamarin meetup page because, for whatever reason, uh youtube wouldn't let us put the links to those in the chat, so it's there, but you can also just search for resilient network connections in xamarin. There's great articles that talk about how to effectively use refit and poly and even akavatch, for you know, caching, your your stuff and on the mvvm stuff, I did put a link in the chat about the microsoft, learn article that covers how to create mvvm uh models uh and it walks you through that whole thing.

C

It's like a 25 minute lesson, awesome.

D

All right and rounding out our xamarin forms architecture. If you do want to have local caching or you want to store data locally well, then you got to have a database, and so I wanted to bring this up because sql there's a library, sqlite net. um If you've been doing. Let's show this. This is just a nuget package, so get trends nuget package.

D

Where are you whoops sqlite net dash pcl, ignore pcl? That was what it was originally named like 10 years ago, when frank krueger made it, but this is going to be the most performant database that you can add the reason I say that is, it doesn't have all the bells and whistles of something like entity framework, but as of today, don't use entity framework in your xamarin apps. uh There's known issues with entity framework, the team's aware of them.

D

I've been working with the team and uh matt soka has also been working with the team internally to help pin those down, but there's uh we'll say, use cases where things like just initializing. The database connection on an entity framework in a xamarin app takes three seconds and if you are pulling data from your database when the app launches well, that means your app launch is going to be delayed by three seconds, which is a lifetime in mobile.

D

You know we want our apps up and running and being used by the user within three seconds and so try to avoid any framework. For now. The team is aware of those there's a github issue. If you google it, you can find it. But this is what I recommend sqlite net dash pcl.

D

It doesn't have all the bells and whistles, but it's going to be the the fastest one for you and to show you this database layer.

D

uh Let me show you this base database first, because I want to point out I'm using poly again now, specifically I'm catching these sql light exceptions, but what I recommend is to use sql lite when you initialize it. There are things like enabling right ahead logging, which will give you a little bit of boost in performance, and so we won't dive too deep into the databases.

D

But I did want to recommend sqlite.net and also when you do things make sure to include poly, because what I found is sometimes with uh when you're saving stuff to a database.

D

Yeah we're we're doing multi-threaded things, but if you're trying to write to the database at the same time, sometimes you'll get the sqlite exceptions where things time out. And so, if we just do things.

D

D

Where's my reference- this should should have done it like this find references. Oh there, we go so any time we're doing settings with the database. I put it in this attempt and retry now the biggest difference here with this poly retry is the number of retries. I crank it up to 12 and I only I do the same exponential back off, but at the millisecond scale, so the first time I'll, wait, two milliseconds, then so on and so forth, because milliseconds is like a lifetime when it comes to cpus and databases.

D

So that's a little different than when we're hitting apis.

D

All right on to testing now there's two kinds of testing for our xamarin apps. There is ui testing which we'll talk about first next up we'll talk about unit but ui testing, emulates a user using your app. So users can tap buttons on the screen they can pull to refresh. They can enter text into text boxes they can navigate to and from pages sorry there's a question sure what about dapper.

D

um No clue I've never used it. Oh so damper.

C

Is like an a very light orm?

C

So, oh! Yes! Yes! Yes, that's right! um I don't know that it wouldn't work the lahi. I don't know if I'm saying that right, um I haven't tried it, but it might work. I don't know why it wouldn't because it seems to work with tottenham.

D

C

Another thing you can look at is also akavatch, uh which is a whole caching framework.

C

C

And it'll, let you manage the lifetime of you know any cached object. It's pretty cool! It's in that same link that uh uh net foundation posted above resilient networking and xamarin.

D

Right out, yeah, no good question, that's right! I I I remember dad for now um yeah, I'm unsure uh how its performance does on xamarin, but you know if it's something where you try it out and it works great go for it. But if you notice things like we've been seeing with any framework core, where there's things like three seconds to do a database connection or a couple seconds to do a database migration.

D

Those are essentially show stoppers for me, which again the team's aware of it, we're looking into it. Hopefully someday we'll be able to do everything in any framework core because yeah, it's great for back-end databases that live on servers, really powerful servers that are always turned on. But uh mobile is this kind of unique, weird edge case that you have to make sure you you plan ahead for it.

D

So yeah, circling back to testing, ui testing, emulates, a user and the way this works in xamarin forms is xamarin. Forms has every ui element in xamarin forms, so a button a view, a label, content page anything that exists in xamarin forms has a property called automation id, and so what I like to do here- and this is where reo is saying there- are places where I do recommend a shared project.

D

uh This is one of those places, because what we want to do with our ui tests is, we want to use automation ids for everything. Let me show you what this looks like real, quick, let's jump over.

D

D

We'll do a find reference, so things like this close button is being used on my referring sites. Page. Oh, that's the ui test code.

D

There we go so when I create this close button. I tell it to use this automation id, and this is my xamarin forms code. Now in my ui test code is the one I open first jump back to, that um we can reference that same id. Automation, ids are just string constants and it's essentially a unique identifier that we can use for our tests. So this close button is using that same string, constant close button that we're using in our xamarin forms code.

D

So the reason, let's see so the reason I recommend having a shared project is well the reason I recommend sharing that code. First is so that you can have these public string constants that are shared by both xamarin forms and the ui tests. I learned this the hard way because I used to just hard code the strings and I kept having typos, so my ui test would fail because I couldn't find the button, but it's because I spelled the word close button wrong in my testing code.

D

So by having a constant string, that's shared between these xamarin forms project and my ui test project. I don't have to worry about typos anymore and the what these ui tests look like. If I want to say things like tap, this store rating request button, you just say app.tap.

D

We pass in that button that we initialized that uses the same constant. The ui test will scan what's visible on the screen for that string, constant and if it finds it, it'll tap it. If it doesn't my test fails. So the reason I recommend using a shared project here is going back to the fact that.net standard libraries generate a dll um if what we're doing by using a shared project is just speeding up our compilation time. So I don't there's two things going on here.

D

I don't want my ui test to reference my xamarin forms project because well a it doesn't have to and b that's going to take forever to compile, because when you click compile it first compiles all the references. So if my ui test project was first referencing, my xamarin forms project.

D

Well, it's got to build this whole dot-net standard library and its dll, and then it can build its own dll and then run the ui tests, which can get a little painful so having a slimmed down separate project, essentially, for these constant strings makes everything really really fast. In fact, it's a shared project. We lose. We get away from the overhead of having to first generate the dll, then compile the head project. So this is one place where we're just using string. Constants.

D

Unlikely that you'll shoot yourself in the foot and we can get a little bit of performance benefit now. The ironic thing is my git trends. App is actually using.net standard.

D

Basically, I'm also sharing I'm sharing a bunch of data in this mobile common project.

D

So these are all my automation, ids, but all that all those translations, all these resex files also are shared so that I can run my ui tests and I can test out the app in spanish and I can test out the app in german and the only way to do that is to share all the information between um xamarin forms project and my ui test project, but resex files, just don't work in shared projects, there's a lot of code generation that happens in the background by visual studio and that just for whatever reason, isn't supported in shared projects.

D

So I had to use it in standard library here, so I'm being a little bit hypocritical, but that's okay, uh because using resex files. So that's my excuse, anyways and then there's also things like we can share certain services like this uh mobile sorting service. So, like the app closed out, but we can sort by views, we can sort by clones we can sort by stars and all that logic I want to be able to verify in my tests.

D

So I put it in my mobile common project, so my ui tests, my unit test can also verify that as well and there's also some models here. um Ui tests have these cool things called backdoor methods where you can literally.

D

Well enter into the back door of the mobile app so instead of just tapping things on the screen, if you want to confirm what's in the list, you can write some backdoor logic into your android app into your ios app and when you do that it has to pass back a model, and so it's easier just to share those models between, so that we don't have to essentially rewrite the same code.

C

All right, we have a question asp nerd asks. I don't recall if you covered this or not, are there good resources for already made styles or style sheets for branding and nicely formatted app parts like the list? Your friend helped you redo.

C

D

I can't think of any off the top of my head um they're.

D

Actually, what do you do? Ari? Because I'm not.

C

Good with you eyes- and so I know that there's so if you haven't seen it, go check out sync fusion, they through a whole set of libraries and as long as your app makes less than one million dollars a year. It's free, so that's kind of cool. I didn't know that, which is why I wasn't using their stuff, but they're free, um I believe telerik and infogistics also have uh ready-made templates. I know there's a number of sites out there. There was even like a xamarin gallery at one point.

C

I don't have the link to it. Let me see if I can find it um uh yeah, there's a was it the xamarin forums, there's a really cool, like xamarin forms like control gallery at one point. That was really I just. I don't know where it is at one point um I don't know where it is anyway, somewhere on the internet. At one point I had it linked, but yeah syncfusion, infrastics uh telerik they they've got that stuff, uh syncfusion's, free and I'll. See.

C

If I can look that stuff up, uh the other thing is is as you get as you build components in your apps that you like, because you will you'll be like wow. You know. I worked really hard on that. um You can extract that relatively easy easily into a reusable view, so try to build your libraries of those reusable views like if you go the xaml route, because I'm a xaml guy.

C

um I find it very easy to have those reusable components and then I just inject them in the xaml and then I can just swap it out. You know, as I see fit or reuse it all over my application, like a header very another one, to take a look at go. Take a look at sharknado.

C

I keep shark nato, sharp nato I'll I'll write these down real quick, so people will forget so sharp nato is one there's like a sharp nato ui. I know uh matt and uh james were talking about some of the cool stuff. It's got with, like the the like the blur background effects and everything. That's really neat there. You go yeah material frame.

D

Yeah, um I just think, what's really ironic, is I'm literally using those in this app.

D

Like this chart is all syncfusion, I didn't write any code for this. I just use the syncfusion library, which is very convenient.

C

Sweet, uh let me write down the other. Someone was kind of a team special job, also uh progress um or probably tulera cat stuff and logistics.

C

um Infusion is free malahi as long as your app makes less than a million dollars. So if it makes a million or more uh then hopefully that means you can afford it, but it really is free. You'll see just uh I can walk you through it. If you ever have a question just ping me, I I'm at re on twitter, but yeah, it's it is free.

C

uh Let's see what else are there any other questions? How does mvvm work if you want to display more than one record from your database on the screen? Doesn't model only describe the data types of one record? I can answer that one, but I'll. Let you do it.

D

Sure, hopefully, we're on the same page all right, but I was gonna say if your view model just has a list of that model. um I mean I do this for all the lists in this app like each of these.

D

uh The data that appears in each of these rows is a repository model, and that contains the data for the name of the repository, the avatar url, the numbers, reviews and stars, and so yeah. You just put it into a list, and then you can use nvvm to essentially extrapolate that view for each row and it'll populate in our in our list like this. Is that what you were thinking too.

C

I wasn't listening. I was looking something up, so yes, I nailed it. I really want once you have data. So if you have your your view model, you know once you once you've got your data, you can display anything. You want the way you want to. You know you're not bound to a particular data type. You can make things as a as complicated or simple as you like. You know, that's that's really up to you.

C

The the view model will hold on to that data and it will, you know, handle the the actions you want to take on.

D

C

But you know how you present it that's up to you. That's actually not the view model's job, the the the view would do that sort of work so.

D

Yeah short answer: yes,.

C

I just posted something about visual, so visual material uh that'll. That may help that's another cool thing. That's out there and I'm trying to find uh you know. Oh here's, the um the xamarin ui kit uh uh templates from sync fusion, all right. I can do this without nice, no problem, I'm always got it. It's so deep.

C

I remember when none of this was free. You know it was all expensive and now it's like yay.

D

Yeah the ecosystem is pretty amazing. That's I will say just a general recommendation. If you are about to do something, um google, it first to see if it already exists like this, um so sharp nato, this uh material frame. This is sharp nato material frame and you can do cool things like um that. Has this little bit of elevation, so it looks like it's popping out of the screen and I do things like if it's disabled, the elevation will go away, so it almost shrinks back into the screen.

D

So it's it's subtle, but you can see that elevation. Isn't there compared to these other ones. So yeah just super cool things, and this is like a couple lines of code because uh who makes sharknado jean-marie did all the heavy lifting for us and he's way better at it than I am so yeah. Definitely just do a quick search to see, like maybe there's like a shadow button that somebody's already made that we can leverage.

D

All right, so circling back to the architecture next up is unit testing. Your your unit test has to reference your in this case xamarin forms project because that's where your services live. That's where your view models live. That's where your database lives and the rule of thumb with unit testing is have a test for every service have a test for every view model, and probably more specifically, um I should say every public api, because this services, each service class, will have protected and private methods and a common question to get is well.

D

How do we test those? Well, you you don't! I mean there are ways to do it, that I don't recommend it'll. You can use reflection to access private methods, but then you're kind of breaking the whole point of c sharp in my opinion and so yeah test every public api.

D

Those public apis will reference those private methods, and so you can still get to them and you just have to make sure you have a test that covers those paths, and so what this looks like in git trends: let's go ahead and shrink that down a little bit. So my unit test all these tests you can see. I have here's, my all my tests for all my services, all my tests. For all my view, models um I even test all. I have this test to make sure we have res x.

D

So if you open up, if you send somebody who knows chinese my way to get trends, we can add that in here and the rest x test make sure that all the press x is valid, and so these are life savers. You know for git trends.

D

This is a free app. I don't make and haven't made any money off of it, and I am, I won't say the only developer working on it, because I've had amazing help from folks in the community, but probably 98 of the code. I write and maintain. So this is just something I do for fun on nights and weekends.

D

I made this as a tool to help me because I saw the need and love that other folks are using it all around the world, but the only way I can survive as an individual developer, an indie dev on a free app is by having all these unit tests. So things like you know all these uh apis, so testing things like um making api requests for valid repos, making api requests when it's unauthenticated, making ipa api requests for invalid repos.

D

Having all these things here, allow me to add new features, and then I can quickly run my suite of unit tests same with my suite of ui tests and know that I didn't break anything and then, of course, every time you add something new add a new unit test for it, and this is really where dependency injection comes into play because, um specifically with xamarin essentials, if you try to use certain xamarin essentials, libraries in your uh unit tests, it just blows up.

D

There's certain samurai essentials apis that just don't account for uh net core, which is I'm using this unit test, library.net core library, and so they just don't work, and so the only way around this is to essentially mock them out, so basically creating about this folder for mock services.

D

So things like the browser, well, dot net core doesn't have a browser, but xamarin essentials does, and so, if I have code in my service layer, my view model layer that opens up a browser and I try to unit test that it's going to blow up it's going to fail, not because my code is bad, but because there's just no implementation for a browser. So there's this amazing library, xamarin essentials interfaces, that's made by my uh good friend ryan davis, and he essentially takes every version of xamarin essentials and then creates an interface around it.

D

So this ibrowser is an interface that we can use to essentially create a mock browser implementation, and all this allows us to do things like well, coupled with dependency ejection.

D

Now there they are so I can create all these mocks so like or analytic service right like this is app center. App center doesn't have a thing for net core uh because it's made for ios and android and then all these uh xamarin essentials interfaces that I can mock out and what this ends up. Looking like is back in that xamarin form source code.

D

For things like I do some background fetch. I can tell xamarin forms dependency injection that I'm just going to pass in an iap info which, in the xamarin forms dependency injection. This will inject into it the xamarin essentials iap info, but in my unit testing it's going to eject in a mock app info that I created.

D

So this is where dependency injection really became critical for me was when I started to uh essentially up leveled to making sure I unit tested everything like a good responsible developer, which I mean really it's only my app store apps have unit tests. uh I have a lot of just playground open source projects that I don't probably should but um yeah.

D

If I'm gonna push it to the app store, it's gotta have a suite of tests because I'm the only developer working on it and I need to make sure I don't break anything or if anybody from the community opens up a pull request that they didn't accidentally introduce any new bugs. So um highly recommend xamarin essentials interfaces, it's just a nuget package, but yeah when you're doing unit testing. This is really where dependency injection come well for me became critical.

D

All right, we're almost there a couple more slides, because the next thing, because that we want to talk about, is including our back end, because xamarin forms is all done in c. Sharp um our unit tests are all in c sharp or ui. Tests are all in c sharp and our back end is also all in c sharp. So we can create apis.

D

We can create azure functions and because everything's written in c sharp, we can share code, and this becomes really critical when we talk about making our own set of apis, because our apis will expect a certain model and our xavier forbes app will also expect a certain model, and so, instead of duplicating code, we can put these into a common folder, and so this is another place where I recommend using a shared library, because this code should all be we'll say application or platform agnostic, because we're just talking about um some models.

D

Some interfaces- uh nothing in here should be anything where we rely on a specific framework. So we can use a shared project and again that make sure we don't create a dll every time we can take advantage of that compiling performance benefit, but again, if you're uncomfortable shared projects, you're not familiar with them.

D

Dotnet standards still totally cool, and what this looks like and get treads, is this little folder right here? So things like refit, so I have. I have azure functions. If, if you've never heard of azure functions, it is azure's serverless implementation, so functions are really cool because they're, basically just a method um that runs in the cloud, it's serverless, so it only when it's not needed. It scales all the way down to zero, which means that, yes, this is an azure.

D

Yes, you have to pay for azure, but with serverless azure functions, I've never had a bill higher than 10 cents a month, so 10 us pennies. I can run this code in the cloud because it only runs on demand, so certain things like creating that login token.

D

Needs to be done in my azure function, but that's also code, I'm leveraging in my mobile app. So I can share that code, even though yes, it's a service, and so I have all the refit implementations, any models that I'm passing back and forth, like specifically, these dtos so like um we are passing back.

D

This generate token dto. So this is what has to be generated on the server side, but it's also what's going to be received on the client side, and so instead of duplicating that code and making sure well duplicating code is always bad, because if we update something on the server side, then we have to remember update on the client side. Instead, we can just share that as a model between the two. So anything we want to share between our cloud back end and our mobile app.

D

We can put in this shared folder here and then also things like constants, like the github apis, the os scopes that get trends needs all this stuff can be.

D

Stored here and then that way, we can share it between our our servers, our apis, our azure functions on the back end with the mobile app, and this really to me is the biggest benefit when it comes to being a c-sharp developer, because yeah there's technically swift servers and kotlin servers, but they aren't very popular. um I can't speak to whether you know their performance or anything cause.

D

I haven't used them, but they're, that's not usually the go-to when you're creating back ends, you don't usually say: oh I'll, make this back and then swift, but with c sharp. We have very robust tools like asp.net, core and azure functions, and all these great things that we can leverage and it's all a.net. So it's a dot-net. We can share code and that way we don't have to worry about maintaining server-side code versus client-side code. We can share that code between.

C

Yeah sure encode we did yeah, that's why we used tamarind anyway.

D

C

I did have a question um from bachelor. um James was saying there were reasons not to have interfaces within xamarin essentials. Any idea what reason, what that reason could be, and thank goodness.

D

Yeah, thank goodness for ryan davis and sam essentials interfaces. I I don't know um my hunch is that well here's! Here's here was my experience. I agreed with james until I needed to do uta testing, um because until then all my xamarin essentials code was running on a xamarin app, so yeah the browser implementation worked every time because xamarin essentials has that implemented for ios and android and great and so um yeah.

D

I'm not sure if it's just a thing where they haven't felt the pain of writing a unit test, but one of the other things that is we'll say a common thread within.

D

Say the.net world is when a good open source library is created.

D

We don't want to overtake that um because that was that's kind of like an uh but basically yeah. We don't want to um step on anybody's toes, and you know ryan made this library he she he should and does get all the credit for doing it, and so that's been a trend. I've noticed where folks will ask that for other xamarin essentials things as well like hey, can we bring this in and the initial answer is typically well. It already exists. This already works great, and so yes.

C

D

Could implement it, we could roll it in we'd have to tweak it to fit into xamarin essentials, um but for now yeah just use this other library. So yeah, I don't know, I don't know if it's maybe something we could get on the roadmap but um yeah. I wish I had an answer because I I wonder the same thing I I wish it was in there. I know. I know the original design decision was to keep it simple, which is why everything was just uh static.

D

So you didn't have to create instances. You didn't have to worry about doing that. So the simplicity is there. But I say why not both good question, though.

D

Okay, so we are way over time, but let's round it out with what is coming down the pike we have one.

C

Question, uh let's see, can you show some code where one of these ids is defined set.

D

Up, let's automation, id.

C

Let's talk on, I think, that's the automation id. uh Let's talk about, let's answer that one after the future architecture, so people can see the rest of your presentation. So if you don't mind mark on that, one.

D

No problem that that'll be a quick answer too so yeah last but not least like what's what's next right, you know we, uh I r and I were chatting about.net maui coming and yes, it is coming it'll, be in dot net six dot net, six, the first release, so the very alpha preview of dot net six should be available to everyone in november 2020.

D

So we're only a couple months away from at least seeing the alphabets alphabets alphabets, but I personally won't be porting my apps to.net maui until we get at least a release candidate out, because the initial preview releases those apis are subject to change. So I don't I'm I'm being selfish here, but I personally don't want to have to port my app to an alpha version of net maui and then have to report it again with if the apis happen to change, but with dot-net maui.

D

Our architectures are going to look very, very, very different. There's a lot of things we have to do in as xamarin developers today that are just a construct of how net works.

D

So things like, if we're in our xamarin forms project, and we need to access an api that exists in our ios project, we have to do things like either dependency injection service locator to do what's called inversion of control and get back into the parent project, and that's not easy like when I was first learning xamarin uh that took me a while to wrap my head around.

C

D

I imagine it's true for other new developers and so what we're doing with dot net maui is. We will be marching towards this world of what's called a single multi-targeted project. Now multi-targeting is a thing it's been around for a couple years, but it's mostly used in libraries. So actually speaking of xamarin essentials, cameron, ascend, joel's github, if we check out the xamarin essential source code, camera essentials is using a multi-targeted library, and so if we look at things like, if we talk about the browser, let's check it out so you'll notice.

D

There's all these things like browser.android.cs, browser.ios.cs and then there's also browser.shared.

D

So the way multi-targeted projects work is if I zoom out and look at the cs, proj what we do. We say target frameworks and then we include the frameworks we want to target so dotnet standard, xamarin.ios tv, os, android, the different flavors of android tizens, everyone mac. So all of these frameworks are targeted in this one project, and your next question is probably well hang on. There's no way you can do that because you're going to blow up the compiler, because there's apis that exist in android that don't exist in ios and you're right.

D

So the way we handle that is by using the file names, so browser.android.cs will only compile when we are compiling for android. Likewise, browser.ios.cs will only compile when we compile the ios project and obviously shared is the one that gets compiled for all of them and then there's also.net standard but anyways. If we look at the shared code, this is um this is using partial classes. So this is the world of multi.

D

Single multi-targeted projects is everything's a partial class, so the shared file has the generic api layer right. So like open async, this is the one we call from our xamarin forms apps when you're using xamarin essentials. I say: oh wait, open async and I pass in the url and then they're doing something under the hood they're calling a different met, a different method, we're just calling a different method and what we'll notice is? There's a lot of methods missing from here and and that's because those methods live in the platform-specific implementation.

D

So by using partial classes, we can have our shared code and our platform-specific code in the same project and then, instead of having all these different projects and using a version of control to get back into the platform-specific apis, we can just rely on the compiler to know. Oh, hey, I'm compiling for ios, let's make sure to include browser.ios.cs, and so that's where we can call platform openasync- and this is has all the sf safari view controllers.

D

All the platform-specific apis for ios can go in this file and we don't have to do any ioc craziness. So the future in.net maui will be to leverage these multi-targeted projects, and the difference here is xamarin. Essentials can do this now. Do this right now, because it's just a nuget package, it doesn't have to worry about, including a runtime.

D

um If you make a new get package, you should be using multi-targeted projects. Well, let me represent, I highly recommend it uh I say should be using because literally five days ago I just ported. My nuka package called async weight, best practices over to single multi-targeted project, but multi-targeted projects has been around for about two years now, so it only took me two years to catch up, don't feel bad if you haven't yet either, but for libraries, yeah multi-targeted all the way, because it makes your code a lot easier to write and maintain.

D

You have to worry about inversion of control, but what we don't have yet and what is coming. In.Net, 6 and maui is the ability to do this with run time. So um libraries, you can do that today for creating a new get package. Multi-Targeted project awesome dot net maui. We can now expand this to things like here in this screenshot. We see well. First of all, we just see one project.

D

There's not my app.ios there's not my app.android there's not myapp.uwp, there's just my app and if we want to access those platforms specific apis, we just create a file called bluetoothadapter.android.cs and we can do whatever we want with the android apis. If we want to hit ios apis bluetoothadapter.ios, we can do whatever we want, and that way we can just use partial classes and this secret sauce in the compiler, with file naming to access all the platform-specific apis for us.

D

So um the other cool things that we're working on is being able to share fonts um font. Embedding is actually already a thing in xamarin forms like I'm doing that in git trends I have fonts in my xamarin forms project here, uh resources fonts, so I have all my fonts here and then I just have some assembly attributes to say export this font and I give it a name, so I can reference it elsewhere in my xamarin forms code.

D

But this way I don't have to add each ttf otf file to my ios project and my android project and make custom renderers for it like we used to have to do now. I can just use this assembly attribute and then in, like my splash screen page, for example, I can just say: font: family roboto, bold, so multi-targeted projects are the future and what that entails is a partial class for it's not shown here but well. We could go back to essentials, so a partial class for yeah.

D

We can do this even better, show them side by side. You go there, you go there.

D

So public static, partial class browser public status, static, partial class browser, and these two can just reference each other as much as they want. It just happens to be that this one is our ios specific apis, and this one is our shared apis, so non-platform specific, and this is how we will be able to take advantage of um multi-targeted projects in maui. I'm really looking forward to this um if, but uh we won't be able to play around with this for another couple months.

D

So one of my plans is when dot maui comes out in november and we get our first peek at it. I'll probably take a simple sample app or dummy app that I've made in the past. That is not in the app store and then try to port that over using these partial classes and the file extensions just to get a feel for this partial class, because I don't know about you, but I never used partial classes. I always thought like.

D

I knew they existed and I know that's how xaml works with it takes advantage of partial classes, and I knew that but like I never saw the need to create a partial class myself, but now I see the light like having partial classes and compiler specific file, extensions or platform specific file, extensions for the compiler. This is just going to be a beautiful world. You know imagine having somebody new, join your xamarin development team, they've, never done.net before you don't have to teach them about inversion control.

D

You don't have to teach them how to access platform-specific apis, because it's as easy as just naming your file with ios.cs and making it a partial class.

D

Okay, that's it ari! That's all I had lined up. Do you have any questions about.

A

This a lot of information, um so we have a couple questions.

C

uh The first one: uh let's see, can you share some code where one of these ids is defined, so I think that's we gotta circle back to that. Yep.

D

So these are just uh and we're also assuming you're talking about or the question is asking about uh automation, ids. So if that's the case great, because I'm about to show you, if not feel free to make another comment to correct us, but for example, um let's see I had that referring sites page, so it's literally just a public, constant string, you can name it whatever you want. I could have named it apple tree, wouldn't make any sense, but um because this constant just needs to be the same it just.

D

Let me rephrase that it needs to be a unique identifier and it needs to be the same. Reference in my xamarin forms app as my ui test app. It could be whatever, um but obviously apple tree wouldn't make since here. So what I do, I'm kind of lazy and I'm also bad at thinking of names so to keep it unique.

D

um I just say name of and reference the the actual constant and the class that it lives in, and I do this I have tons of tabs open. Let's do this, uh don't save anything so I'll. Show you one side by side like the repository page.

D

Also has a collection view so since these both have collection views in them to keep them unique, that's why I use the name of the class that resides it plus the name of the automation id, and so when I reference auto repository page automation, ids.collectionview, uh my ui test knows to look for this string. That is repository page automation, ids collection view. So this is just the way I do it like. I said you can just you can always just do reflection view one two three and that's fine. It just has to be unique.

D

I guess you could also do a no don't do a good, I'm not gonna. Go there, do a hard-coded string or leverage name of if you want to as well ends what the good stands for. What's that.

C

I was making a joke every time. Every time I see something in the in the in code. That says, I don't know whatever there's not even a variable name, but just a guide is in there and no one knows what it means you know. Do it come from? What does it stand for? Well, didn't you know? That's our application, id and azure. You know azure ad well, instead of variable name for that.

C

um So we had a question um if you have an external set of apis that you manage. What is your method of authentication from the mobile app into that and how do you handle new users etc?.

D

That's a whole other meetup.

C

So asp nerd- and I also pointed him to the xamarin essentials web authenticator, because azure ad, you know, there's a whole authentication system for that and another thing that you might want to take a look at asp. Nerd is um azure's. Api management and azure has or probably amazon aws has this as well. It'll actually take any number of apis that you've got and it'll abstract them put them behind a single set of security.

C

So you can control the flow from different colors or different security, all in a ui, so you can take a bunch of uis and put them behind one api management structure.

D

C

I'll show you he added a follow-up question for that. I think, if you have your own managed api, how do you handle your mobile app connecting to it for security and how also? How do you handle users in your app.

D

Yeah, so let me we'll break this into a couple parts, so if you're using azure azure actually makes it pretty easy to implement authentication, I just want to show you this, like I was telling ra. I created my first website just this week. Like I said I never was a web developer before getting into mobile, and if I try to navigate to this website, it should ask me for my microsoft credentials. This is just an internal tool, that'll be used by my there. We go by some microsoft employees and so yeah.

D

It's asking me to log in with my microsoft credentials and what's cool with azure, I can basically just toggle this so.

D

Let me click into it. There we go so this authentication authorization, it's turned on and specifically I'm using active directory. If you wanted to, you could do a social auth or something like that, but check this out. If I just turn this off, save it and ah shoot, I forget the url for my website. Let me launch the website again and.

C

Go back a couple hold down the uh the back arrow.

D

Or whatever you got it so navigate to the website again,.

D

Should let me write in and there's the website, so uh this was actually pretty easy to do, uh just because we have this little toggle and azure, but um this requires uh you to have an account already. So that's one of the also one of the reasons why it's super easy.

D

uh Let me show you what I'm doing in get trends, though, because, let's see just, let me think ahead, real quick to make sure I don't give away any api keys, yeah, okay, so in this app, so I'm using azure functions, serverless functions for my apis, so things like certain things need to be authenticated. Certain things don't so things like.

D

I keep a url for there's a little intro video when you first launch the app that shows you, basically just how to use the chart like how you can tap and drag or how you can zoom in and out to see more data, just kind of like a little onboarding experience, and I kept I keep this in.

D

I keep the url for this stored in the cloud, because if I want to change the video, then I can just swap this out in azure and I don't have to release a whole new version of the app if it only if the url was hardcoded in the app, I would have to release a whole new version of the app, but um so this is just anonymous. So anybody can hit this api and get this video url.

D

I don't know why you'd want to do that, but you're welcome to use my video anytime. You want because it's an honest api there's certain things that I want to keep hidden and azure has the or specific azure functions. Allow me to increase the authorization level and so to do this one. This one is what I use to pass in the syncfusion api key, because I do have a paid syncfusion account and when you have a paid sync fusion account, they give you a special key that you have to load into your syncfusion controls.

D

uh Otherwise, let's see I don't know if we caught it earlier.

B

D

Come on android force quit there we go um because if you launch the app without that api key you'll get this warning, we go get past the splash screen, boop and I'll. Give you a little heads up. You might get a preview warning, but um if I click on one of these there we go syncfusion. This is kind of their their way to make sure people are using or paying for the service right. So um I keep these api keys again stored in the cloud, so I can rotate them.

D

If I have to like, if an api key gets compromised, I want to be able to rotate it and again, if that was hardcoded into the app, then I couldn't plus, if I had coded into the app well, this app is totally open source. So then anybody would have access to my api keys so uh yeah. This works by um essentially blocking anybody that hits this api. Without a specific api key. You can't get my secret top secret uh syncfusion apis.

D

So I mean I I I joked earlier, but I said that's a topic for a whole. Another uh meetup aria. Actually, I'm happy to come back and we can chat more and dive into it deeper, but yeah. That's always authentication is really tough and I think one of the things I'm fortunate with in this app is when you log in you're logging into your github account, and so I didn't have to actually create a authentication server, because github's doing that for me- and I just follow the oauth flow.

D

Oauth is kind of this authorization standard, and so I just basically tell github, I'm like hey um this user wants to log in with their account and github goes: okay asks the user for their credentials, you'll actually notice. My credentials are now cached, but you'll notice. It launches the browser.

D

um Oauth always will launch a browser, because that prevents me as the app developer from ever seeing your credentials. So certain developers don't like that. Oauth launches the browser and they will uh do what's called an implicit workflow. Don't do that because that, if that app developer is malicious, then they can have code that logs your password to wherever and then they just got your login info.

D

So yeah oauth launches the browser there's just basically, I just send the user over to github ask github to ask them to authenticate with github and then what github does github just tells me as a mobile, app developer that says yep. This is brandon minic or yep. This is ari, and then it gives me an authentication token. So I never see your username and password oauth kind of handles that for me, and so that's one way to do it.

D

uh You could always have a social login like github, twitter, facebook, google, somewhere some service that handles authentication for you or in the case of that website we were showing earlier. You can use azure ad, which just want to make sure I turned it back on so good question yeah all right. Maybe we can queue that up for a future meetup, because that is a very complex topic. That indications are all sorts of answers.

C

Now, to be fair, uh if you don't want to use social login and you don't want to use, you know, federated logins or azure active directory and that stuff which those are cool, um they've got a learning curve, but you know there's a lot of libraries that will make it really easy, including xamarin essentials and it's a web authenticator.

C

That's just built in takes care of all this stuff for you, but let's say you've got uh just using asp.net, not asp.net core, but just you know traditional asp.net with web api and it builds that whole uh web. They call it um owen, you know, along with uh bearer tokens and so forth. um Refit is very easy for handling that. So you know you can use refit to just tie it to the token endpoint of the the asp.net web api site that gets built for you.

C

That comes with authorization and authentication provider out of the box. It's all taken care of for you now when it comes to managing users. That's a different thing. There's examples that handle that for you, that also will sometimes come out of the box. I think at one point there was something there, but then at one point there wasn't, I don't remember. I ended up just writing my own stuff um and the framework has its own identity uh context uh that you know will build all that.

C

You know it'll, build the asp.net users roles and everything in your database and take care of all that for you uh and if you just take a look at, you know how that works in the traditional asp.net uh web api default project, you'll you'll be in good shape there. If you just need something simple yeah, I think the best advice I can give you.

C

I don't know why asp.net corey didn't have that. I think you have to wire that up yourself. I could be wrong. I never saw it like working great, but then again I haven't played with it enough to say.

D

Yeah, I think the the best advice I can give anybody that is new to user management. Authentication authorization. Don't you don't create your own tool? um This is very complex stuff. I mean I use it, but I'm not an expert in it and so make sure you leverage the existing tools. Like I already mentioned, you know, editing framework has tools built in azure. Has tools built in aws? Has tools built in um leverage the tools made by the experts?

D

I mean if you are a security expert, and you know everything about this, go for it, but I'm not, and this is something that is quite literally. What gets most companies burned like when you hear about like data breaches?

D

uh Probably the biggest attack is a phishing attack and then it comes to just like you never change the default password on your user database like it was password password or admin admin like yeah. That's how then like companies like well, I won't throw any company under the bus, but major corporations leak, all this user data and it's things and it's our data. It's like my credit card info.

D

My address all my private data that I trusted you with is now gone because you tried to roll your own service, so yeah, don't don't try to roll your own service um stand on the shoulders giants, especially when it comes to security.

C

Yeah they have the budget to protect you. If anybody remembers that the original version of the sql server the default sa password was nothing, it was empty, it was. There was no so oh gosh yeah. That was the bill. What's sql seven and I think sql 2000 it wasn't until like people 2005 or sql 2000. it no sql 2008. I think it was sql 2008 is when um the you weren't allowed to set a blank default password without, like some ui ui warning going whoa, it's terrifying.

C

Who would ever have access to the database server on the.

D

Network good question uh anything about net maui or multi-targeted project. Sorry,.

C

uh I haven't seen any questions come through. I'm excited about the multi-targeting, uh especially what looks like a ui that makes it not suck um you know. So I didn't really know anything about the multi-targeting until tonight. Actually, like I had heard it, but but no one ever pointed out to me like here's, how it's structured and um now I'm like. Oh yeah, it's one of those things you kind of have to know how to set up the cs.

C

Proj, you know, and then you know blah blah blah, but it looks like maui will bring that over into the ui, uh which would be wonderful. Oh here we got a question. Do you have a good app testing for security to check your app?

C

I I don't know I don't think, there's an app that really tests the security of your application. I think there's, uh if you haven't checked it out, go check out the owasp. um You know what is it the the application security best practices? I I gotta remember top 10 top oh wasp um best practices.

C

Definitely look at that dave says excellent presentation.

D

Brandon. Thank you. Dave thanks for sticking around.

C

D

C

D

A one-hour meet-up lasted a half.

B

It is what it is yay. You know you can't get in the way.

C

Of knowledge, um but you know, if you're concerned, I mean follow the best practices, for you know: application development right there, what brandon's, showing you, those top 10 web application security, risks those to a good extent, also lean into mobile. You can also build you know, there's tests that you could build if you're part of a corporation there's, actually firms that will do testing. You know security, testing penetration testing against your application.

C

um You know it's it's not necessarily cheap, but there's some best practices. You know, for example, not sending things in plain text and non-ssl.

C

You know um ios, won't even let you do uh non-ssl by default anymore, so yeah android, too yeah android uh is android requiring that now I mean I just I was like gosh. If yeah, I think here's hd 10 https, you got to ask yourself why, in this day and age is 2020, why? Why are you not at least using ssl um other things that you can do to protect your data using certificates for your api authentication? You know that you can revoke azure key vault can come in handy there.

C

um You know not storing credentials in the application. If you do store credentials, don't put.

B

C

In source code store a hash, never never uh check in your code to a public repo, it has your credentials in it again. Things like key vault will help you there, because you can keep your secrets out in the cloud and when you build it'll automatically inject the secrets into the build process, never keep them in the cloud.

C

uh I think there is a um what there there's there's services out there that what they do is they just go through all the public, github repos and pull credentials out and go attack sites, and I think there might even be a public, shaming site that does it too um yeah.

C

Some of this is common sense, um but you know what I think it's sometimes it's not common sense, so you something's happened or or someone who knows better than you, especially if you're new to this. uh This is where code reviews become a very good thing, it's good to get a sanity check from other developers, even if you're a senior engineer or an architect, etc.

C

You tend to be too close to your own code, so there's things that you'll miss because to you everything's great and yeah. It makes sense to go and say, hey and so-and-so or another so-and-so that you trust that there's credentials in there. Would you here's a pull request comment on my code.

C

Let me see if I can fix it yeah. I think that kind of collaboration with other developers can help um and just like you're you're asking here, you know what would best practices be. You know what solutions are out there um yeah ask people like us. I think that entire thing is also another entire presentation secure your app writing secure code, there's. Actually a book from microsoft called uh writing secure code, um it's a thick book uh and it talks all about all the vulnerabilities it talks about buffer overflows. It talks about sql injection.

C

It talks about um uh replay attacks. It's uh there's, you know, there's I don't know it wasn't replay text, but it was like a xss right. Yeah xss attacks, um there's a lot there, but that top 10 o wasp. Oh it's so important.

C

Like there's a reason, number one is injection like if, if you're building an api, don't trust anything coming from the web at all always sanity check your data check yeah, never assume that it's clean everyone wants to attack, you hey, not everyone, but it just takes the one person to get through so code defensively, especially if it's going to be an api or a public spacing where you don't control it.

C

D

Other questions.

C

From anyone and uh asp nerd I hesitate to go and recommend anyone just because that's not really my forte, you know I. I know that I just worked with a company um called dekra. They do it.

B

You know there's there's others that are out there. uh I I I'm not saying that decker's going off, there's a bunch of them.

C

I mean there's consulting terms all over the place, um but yeah you should really just go. uh First start with the the less expensive approach of reviewing your own code and going is it do I feel this is secure. You know like if I wanted to break into my system. uh Could I get past this uh and I think you know like if you immediately go? Oh yeah, I can just do this, that's a hole! You need to pass.

B

Definitely yeah uh any other questions, anyone anyone, mueller mueller and now we have to wait the 15 second delay, because we said it.

D

That's right I mean, if you do, uh I included my twitter handle on the stream here so feel free to uh shoot me a tweet send me a dm uh always happy to chat about salmon stuff, c-sharp, stuff, dot, net stuff.

C

Awesome brandon. Thank you so much for your time, uh we'll make sure that this video gets uh uploaded and I know actually, I think, you're presenting in another couple months for indie, xamarin and then another couple months after that, so hey. uh So, if you check out the indie xamarin site, we've got all those uh presentations listed very excited to have had you tonight and especially for uh future sessions. Thank you. Brandon.

D

Thanks again, ari.

C

Yay and now I don't know how we actually stop the event.

C

You can't see me bye, thank you. All have a wonderful night.