mergeconflict380-1
===

[00:00:00] James: Welcome back everyone to another Merge Conflict. We are here streaming this time in a recording, so we're not streaming live, but we're recording on StreamYard instead of Zencast, which we normally use because Frank is in the middle of a storm in the Pacific Northwest and his Starlink doesn't have a direct line of sight.

So we've bumped down our resolution to 480P and a total output resolution of 720P. And Frank, you holding in there? 

[00:00:26] Frank: Yes, the Salish Sea is rough tonight. Um, the waters are coming from the sky instead of the ground like they're supposed to. All the, all the things are bad for magic satellite internet. So, you know, one of these days I will get it fixed, but I think I've been saying that for about a year now.

So we'll see. Stay tuned. 

[00:00:45] James: We'll see, we're going to see how this works and if it works out well, if it doesn't work out, trying some new technologies, people, so we'll see how it goes, but if this is your first time with us, well, welcome to Merge Conflict. This is a podcast where we talk about technology, development, all sorts of things, uh, usually in the world of NET, usually for mobile development and desktop development with NET, but not always, but sometimes.

And today we have special episode 380. We've been going for way too long on this podcast and every 10 episodes, if it ends in a zero it's lightning topics time. And Frank, uh, you have the first topic, which is all about AI and machine learning. What'd you, what'd you know, uh, what you got for us, Frank? Yeah, 

[00:01:25] Frank: yeah, well, you know, I, I have to get my machine learning into every episode, but, uh, I wanted to talk about something that I'm pretty late to.

It was told, I was told about it a while ago, but, um, Lambda. Oh, I keep calling it Lambda. I should release something called that LLAMA. C What this is, is a way to execute the LLAMA large language model, an AI that you can download from Facebook. Facebook trained it. Um, I find it really interesting. I want to talk about it because, well, it's a convenient way.

So, hey, if you've never run one of these language models on your own computer, you should give it a shot. Um, but what I like a lot about it is the worst part about running neural networks, James, by far, is getting your computer configured and set up for it. It's the worst. Yeah, if you're, if the model's not an ONNX model or a Core ML model, even then you got to like build an app around those and things like that.

Anyway, getting your environment set up is so difficult. Uh, so a brilliant person, and look at me, I'm not, I'm not doing well with my notes here. Uh, uh, what's. What's their name? I will find it. Uh, G, G, G, E, R, Ganov, Gurganov. Let's go with Gurganov. Uh, wrote the fantastic Llama. C which is just a C, C implementation of all the code you need to run one of these models locally on your machine.

And even better than that... You, you see these from time to time, but for the first time, someone did a Mac first thing. Like, usually this is Linux first, CUDA first, NVIDIA first. It was, and it works on all those, of course. Um, but this was Apple Metal first, uh, hacked together in an evening, you know, they say in the readme.

Um, so I just think it's super cool because A, it's fun to run these networks locally on your own computer because they're cheaper that way. And B, I just love the passion of the developer to create no dependencies. Just here's the code, compile the code, and it just. And I think it's lovely. That's it. I'm just so in love with it that I just wanted to talk about it.

[00:03:50] James: So what would I do with this thing, Frank? I just, I can run it as, or like a GUI on top of it that I can like ask it questions or something basically. Or what, I mean, what's the problem that it solves, right? It's like,

if I can download the model from FATE, from Meta, Meta, MetaLlama. Like, I know you just described it, but like, pretend that I don't know anything about LLMs. And like, I download this, what, what am I doing with it? Right. 

[00:04:18] Frank: Right. Um, well, I won't talk about what you would do if you got it from Meta, we'll just talk about what you do with, um, Llama.

C here. So, the README is a little bit long, but it's well written, so you can totally just follow along with that. But, um, with a neural network, you have... The neural network, its architecture, its topology, and then all of its trained weights. And that's this big heavy file, and it's called the model file, and it's usually gigabytes because it's huge.

And so you download one of these model files, and you can go get those from a variety of places. I recommend Hugging Face. If you search for, um, Llama, or Mistral, or WizardLM, there's lots of language models you can get from there. And then, um, this program, Llama. C it's an execution engine. So after you download a model, it's able to actually run the model in a practical way.

You give it text, it outputs text, like the chat interface. So what can it do? Um, it can present like a command line. Chat interface. If you're a nerd like me, that's good enough. You just learn how to run their main program. You point it at the model file that you downloaded and you run an interactive mode and it's a little command line chat program where you're talking to a large language model that you are running on your local machine, not sending anything over the network.

It's all private. You can put corporate data in there. You can tell it all your wishes and desires, and you don't have to worry about someone at Meta or Microsoft reading it. But it's also, it's been super engineered at this point. There's also a server there, so if you want to serve an API, it can do it that way also.

So set it up that way. It integrates with front ends that people have written. Um, the chat UI, if you think about it, it kind of doesn't matter what... Neural network is running in the background. The UI is all kind of the same. So people have standardized the chat UI. I forget what it's called, but it's like web chat dash UI or something like that.

It integrates with that. And it has a million other little integration points. Um, so what started out as a night... This hack to make a tight little Mac app has turned into almost its own ecosystem, and the joke is, like, a whole new model format has come out of it, because that's all we need in the neural network space, a whole new model format.

And it's a little bit funny, because it's a little bit of a failure of the ONNX file format, it's a bit of a failure of the Core ML. File format that, um, a tool like this kind of had to be written, but then they're establishing their own format. Again, it's the wild, wild West. Who knows either way. I think this is pretty special, 

[00:07:10] James: right?

That's cool. And, uh, before we get off of this topic, there's a lot of supportive models, not just LLAMA, but LLAMA2 and GPT, you know, for all. And, um, there's like Refract and Bloom, but there's also these bindings to this library, by the way. Cause it was the next question I was gonna ask. There's actually.

Uh, LlamaSharp. So there is a NET and C sharp, a version of this, and then there's UIs built on top of it. So if you're not a command line person like myself, there's a, um, a small person, uh, uh, out there, a little, a little, a little hacker, little, little small hacker that you may have heard of with the GitHub username of Nat.

Nat. Hmm. 

[00:07:51] Frank: Who could, who could ever get such a short GitHub username? 

[00:07:53] James: Yes. Last name, Freedman, uh, who co founded Xamarin with Miguel and Joseph. Uh, Open Playground is an LLM playground that runs on top of that, apparently. So it's kind of cool. Like good old Nat just coding and working on stuff. Hey, Nat, if you're listening to the podcast, which you aren't, but hello.

But yeah, really cool that this LlamaSharp thing exists as well. That's really neat. And they're apparently going to work on NET MAUI integration. So that was pretty cool. 

[00:08:19] Frank: Well, that's the cool thing. Like, okay. Like. NET's cool because it's cross platform and all that stuff, but the other way to write cross platform is plain old boring C.

And if you write it in plain old boring C, the beauty there is a lot of different languages can bind to you, and NET is great at binding to C libraries, so it was kind of pretty simple to get it up and running with NET. And so I'm hoping to actually, I'm hoping to actually get it running on iOS. So yeah, that MAUI kind of binding and things like that, that would be lovely to be able to run these right 

[00:08:53] James: on the device.

That'd be cool. Uh, all right. Onto the next topic. We try to switch up every five to six minutes or so. Uh, I just want to do a little PSA. We were talking about this privately. Um, but, uh, last week I think we were talking about it privately, but today as we're recording 10. 10, uh, NET 8. 0. Release candidate two came out and kitty.

And with that specifically, um, was the NET MAUI and NET 8 RC2 and our good friend, David Ortenow snuck a little bonus. Well, inside of this puppy here, um, bonus number two, not even number one. Bonus number two, which is for our Xamarin and Xamarin Forms developers. I put a PSA out on my YouTube shorts. But, uh, the team over there, uh, on the Xamarin team, they officially.

Put out a release that Xamarin developers can now use to build officially against Xcode 15 and target the latest versions, iOS 17 and iPadOS 17, and also build for Android API 34. Now this will ensure that you're compliant with the app store policies. Now you didn't need to compile against these necessarily.

There was ways of. It's still going, it's Xcode 14 and even against 15 with some weird stuff Apple removed, but this is now an official release. Now, that being said, there are no new APIs. They do not add, provide a new platform APIs in these releases. Just kind of continuing to enable you to build your existing apps because, you know, in the official support policy, it was never going to support these things, but now it, now it does, uh, in this regard.

And, uh, of course you can, uh, get your migrations going. But if you're, if you're like a few people out there that I've talked to, um, you may be either in the middle of a, of a, of a migration or wanting to, uh, keep it going a little bit longer and you have till May 1st, but this will keep you going for a little bit longer, even technically after end of support.

So nice. Good work team. 

[00:10:53] Frank: Yes. Uh, well, I'm indebted to them because I'm one of those people who has yet to migrate. Uh, even though we've talked about it over and over on this podcast, I have yet to migrate all my apps on over to the new Brave New World, you know. If it ain't broke, don't fix it. But turned out, one of my apps was broke.

Continuous was crashing on iOS 17. And it was all because Apple removed NewsstandKit. Oh, Apple. And just because of the way Continuous works, it linked to every library, just in case you wanted to program that library inside Continuous. And it had a hard link. It could have been, it should have been a soft link or whatever they call those, a weak link.

I forget what they're called, but it was a hard link. The newsstand kit. And so I had to get a release out. That code is a little bit scary for porting over to the Brave New World, so I haven't done it. And so I was very thankful. To find out that Xcode 15, in fact, did work with actually old Xamarin, but they had to do a little bit of a patch again around the stupid newsstand to make it all work, and so NET Thank you.

Thank you, team. Um, I don't know. I don't know who put the effort in, but thank you for supporting 15 and giving me, you know, maybe another year to procrastinate. I'm not going to. I'm totally going to convert these apps, but probably another year of procrastination. 

[00:12:25] James: Yeah, no, I think this is good. And, and, and I did see some people have this issue because they would like install and they'd have like the linker off or whatever.

And that was one of the big problems. Like you usually never have the linker off, but in your case you do. And, uh, specifically if for some reason you're using some crazy library and you're like, I got to just release this app, I don't care that it's ginormous. I'm just going to release it. And actually, you know what, even with the linker off your apps probably actually aren't that big.

I was just updating. What was it? Like some... Oh, they're huge. Come on, man. They're, they're, well, I'm like downloading like Gmail. It's like 700 megs. I'm like, what? Oh, 

[00:13:02] Frank: that's true. Okay. So I'll tell you, I, um, so it's funny. I actually don't have the linker off. What I have is the largest linker XML file you've ever seen.

Yeah. So I actually do green light every library. It's just, I green light every library. Except for NewstandKit now and a few others because, um, like home kit, you have to be approved for that car thing like that. So there are certain things I have to remove. And the app comes in around 360 MB. Not bad.

You're right! You can include All of Xamarin, probably 10 more libraries, 

[00:13:39] James: all of Mono, uh, 

[00:13:41] Frank: the compilers, that's the C sharp compiler and the F sharp compiler, all for 360 megabytes. Now that you've said that, yeah, that's about half the size of a React app. 

[00:13:51] James: So that's pretty cool. It's true. It is wild. Yeah. It is kind of bananas when you think about that.

And people are like, you know, get mad, but I'm like, I want to go. I went to go do an update in my, in the app store. It was like 13 new apps. And I went through those 13 apps where I think I calculated all those like six and a half gigs. Oh gosh. It's wild. Oh 

[00:14:11] Frank: gosh. Yeah. My internet can't handle 

[00:14:13] James: that. No, that's that is wild to me.

[00:14:18] Frank: Apple still yells at you. Like every time you upload a binary above what, 200 megabytes, whatever the cellular limit is these days. Uh, they still complain. They're like, can you get your app up below 200 megabytes? I'm like, no, not going to happen now. The big downside. Don't turn your linker off. It takes 37 minutes to build this app in CI, so don't turn the linker off people.

It's terrible. 

[00:14:43] James: Uh, all right, up to the next topic, talking about Apple development. Uh, we did get a update a while ago from you about trying out, and we did a whole podcast on Vision Pro, uh, development, but there's been some new revelations in the Vision Pro. Uh, drama, drama Saga. No drama. Ss drama. Drama 

[00:15:02] Frank: saga.

Drama saga, drama. Beta, beta time and everything. It's beta chaos, beta chaos time. Um, so while all the real products have left beta chaos time, the vision pro remains excellently deep into. And what happened was I updated my Xcode up to 15 for the release version, and I noticed, oh, I can't seem to create a Vision Pro project anymore.

I can't load my Vision Pro. That's sad. I was having fun making really terrible apps for the Vision Pro. And I wanted to continue doing that. And so I, I, um, finally I went to the website and actually read the docs. And. You have to get the 15. 1 beta of Xcode, so back to the betas if you want to do Vision Pro development.

But then there's a big, bold text right next to it that also says, If one wishes to do Vision Pro development, one must have an Apple. Silicon processor based Mac. And that's new to me because you know what? I'm still rocking James, a 27 iMac Pro and Intel Xeon Galore. And I just got booted off of the Vision Pro development train and I'm a wee bit sad.

[00:16:20] James: Can't you put it on your little Mac? We got MacBook, uh, Air, 

[00:16:23] Frank: right? I do. I do. I got my little MacBook Air. I got my little M1, which honestly is a pretty decent dev machine. But like, it's off to the side. It's not on my desk. My desk is a big monitor in front of me. It's the big iMac in front of me. What I, should I go to like a cafe and drink coffee and develop all my Vision Pro apps in a cafe?

I think so. No, it's Vision Pro. You need like a second monitor just to put the simulator on. And so all of this is just making me realize that it might be time where I'm at the six year ish mark for this iMac, but maybe, maybe for the Christmas time. Uh, we'll have to see. I have to check when Macs are coming out, but I'm, I might start shopping for a Mac because Apple's definitely putting some writing on the walls here.

Um, this is the first time I've seen One of their SDKs require an M1, who knows when the other SDKs are going to start requiring it. What do you think about all 

[00:17:26] James: this? You know, that is kind of wild only because the Apple Silicon has not been out that long. We're not talking about five, six, seven years and those Macs not being supported.

You know, there's. Even like Windows 10 is out of cycle now, there's plenty of support still happening and you can still pretty much develop anything on Windows 10 and Windows 11 and Windows 11 has been out for quite a long time and obviously you can do it like on any hardware. So I think that that, I think it's what Apple wants to do.

It makes a lot of sense because the rest of the product line has finished. Right. I think this is the big distinguishing factor is. All of the Apple hardware is on, um, Apple Silicon. So like, Hey, we only ship Apple Silicon and that's it, you know, at this point. So we're all in that being said, it feels a little premature, in my opinion, to push it off unless there's some technical reason and that they're doing it for, I'm not positive, but there must be, maybe.

[00:18:36] Frank: I love guessing. Can I guess? I just want to guess. Go for it. Um, I think even on this podcast, I've complained about the Vision Pro Simulator. Um, it was crashy. It was buggy. Textures would just... Not Texture, James. You know what I'm talking about, those graphics bugs. Yeah. Yeah. Uh, it had all the graphics bugs you've ever seen.

It was slow. I remember, um, the app I was running was running at 5 frames per second in the simulator. And I started to wonder after this announcement, I'm like, I wonder if this is, like... A metal thing. Um, you know, optimized drivers. And then I got thinking, oh, maybe it's like a unified memory thing. Because the cool thing about the Macs, the M1 Macs you go buy right now is, your RAM and the video RAM are all the same thing, for better for worse.

It'd be nice if the video RAM were faster, but all your RAM, all your RAM is one. Which is really cool because you don't have to synchronize buffers between the GPU and the CPU at all, it's just, it's just memory, man. And so, that, that, that's my pet theory. Um, so, it might not be a herald of the times, it might just be this one unique case of them needing...

Lot of graphics performance in order to run the simulator. Uh, then again, like you said, it really could just be Apple doing their thing of we don't ship Intel's anymore. Why, why are you still running an Intel? 

[00:20:06] James: Yeah, that makes some sense to me. Still hard to swallow. Hard pill to swallow, Frank. 

[00:20:14] Frank: Wow. It's like a 4, 000.

$5,000 pill to swallow if you wanna upgrade your Mac these days, 'cause do you know how many M processors they ship now? Man, I, I, I'm not sure if you can still get an M one, but there's an M one, M one Pro M two, M two Pro M two max, M two Ultra, and I'm like, should I just wait for an M three ? So I really don't know what to do.

Um, If I were to buy one today, I think I'd go M2 max, get as much RAM on it. If you want the ridiculous amounts of RAM, you have to get the ultra, but that gets very expensive. So that's just long way of me saying I spent a few hours today on the Apple store. 

[00:21:02] James: Nice. That's what you have to do. Unfortunately, that's sort of just how the cookie crumbles at the end of the day, because you just have to dive through every single.

Possibility. And then if you're going to buy the top of the line, you really want to buy the top of the line, you know what I mean? Yeah. 

[00:21:17] Frank: And the sad thing is, um, it doesn't look like they're making a powerful iMac right now, so you really have to get like the studio or a pro or a mini. Those are kind of your options.

And then you buy the monitor separate. It's a little sad. I always liked my powerful iMacs, but it looks like I might have to go studio monitor, which is 1, 600 all by itself. And then some expensive Mac on top of that. 

[00:21:44] James: I don't like any of that, basically. That's what I think, but you know what I do like, Frank?

Our good friends over at Google had a great developer event. And by developer event, I mean. Not a developer event at all. Actually, it was a hardware day and we didn't talk about it at all because it happened the day after we recorded. And at this point, I really don't think there's much to really talk about.

Like, I don't know. Seems fine. New pixel devices, new watches, fantastic new buds, fantastic AI, just so much AI in this talk. But I do want to talk about one specific topic, talking about hardware lifecycle, talking about software lifecycle. Um, Google starting with new pixel devices for the first time. In history is providing seven years of operating system updates and feature drops as they like to call them.

[00:22:37] Frank: Feature drops as in they're dropping features. This is Google, right? 

[00:22:42] James: Yes. 

[00:22:45] Frank: No, um, I, I think this is pretty, pretty awesome. I'm, I'm really curious if they're going to follow through. I think the entire internet is with me and we're all just curious, like, Google, that, A, awesome, B, do we trust you to follow through?

So you have these like funny mixed emotions because, you know, fool me once, shame on you, but then, you know, fool me 30 times, shame on me. So I, I, I think overall, I'm like. Dang, that's impressive, because you don't really get that out of Apple these days either, do you? Um, I was just, I bumped up the minimum versions of my apps again recently, and it was all because you just can't get simulators for some of the old platforms anymore, so Apple's being very aggressive too, I think.

Is it 13 or 15? I forget what the oldest simulators you can get these days with Xcode, and that doesn't... That only puts you back about five ish years for device support. And so anyway, I'm just doing those numbers off the top of my head. I think the seven years from Google is very impressive as an announcement, and we're all going to hold our breath to see if they fall through.

[00:23:57] James: Yeah, I totally agree. One thing that I noticed, you know, and to your point is I feel as though I do have an iPhone SE. Verse version or se, uh, se the first se, yeah. The first se, uh, which I guess I only got three years ago, but I think it's four years old now. This did get iOS 17. So that's really nice. I appreciate that.

Nice. Yeah. Now my Apple watch series. Three, which I think they still sell. 

[00:24:29] Frank: Is it okay if I just laugh every time you talk about your Apple Watch 3? Sorry. 

[00:24:34] James: It's, I think they officially don't sell it anymore, by the way. They have the SE, but the Apple Watch 3, which is what I had because I sold it forever, it didn't even get watchOS 9 and they're on watchOS 10 now.

So it's been two years now that I haven't gotten it. So that's quite a long time, but it makes sense that it does get security patches. And I'll tell you this much. Uh, I am on 8. 8. 1 and Heather was on 8. 7 dot something. And I think they had a major fix in 8. 8 because, uh, Heather and I've had this issue for a long time, which is our Apple watch is always running out of storage.

All the time. It's just constantly running out of storage. It's only has eight gigs. That's, 

[00:25:14] Frank: what in the world are you putting on there? I have never, A, contemplated how much storage my watch has, B, run out of it. 

[00:25:21] James: All the time. I don't have anything on my watch. We've uninstalled every app. We have like nothing on it.

It's, it's just random junk in the operating system. And that's it. Wow. And we did notice that the Apple Fitness would download like music and podcasts for the auto thing. We turned that off. That was the first thing. But then hers filled up again. And. And now she's like, she's like, well, I go to my phone and I go to my phone.

I go into the settings and she's like, she's like, well, what are you at? And I haven't seen that in a while. I'm like, I'm at 4. 5 gigs out of eight. So that's so much, so much room left. Right. And, and I don't know what happened. Cause before I was at like 7. 8 or whatever, out of eight. She went to hers and she was at 7.

8 out of eight and kept giving me the pop up. I'm like, okay, try to do your update. She did the update. Now I don't think it just deleted the operating system update. Cause we've been running these issues for a while. Hers went down all the way to 4. 1 gig or whatever. So I think, I think they like cleansed watchOS 3 in 8.

8. 1. So you no longer get the pop up. I don't know what they did, but they fixed it at least. So 

[00:26:27] Frank: that's pretty good. Okay, I don't want to derail this conversation, but it's funny talking about all this storage stuff because Apple's finally put in like the little, um, they'll break down roughly where all your storage is allocated, but there's this big mysterious one called like system data.

Yeah. And mine was like 360 gigabytes. I'm like, What is that? And like, it's not the operating system. That's a whole nother like, you know, 40 gigabytes or whatever, but it's like, what, what is it? And it turns out it's just like old caches, directories from apps and things like that. And there was just no safe way to clean them out.

You just had to go kind of one by one. And I did it. And I recovered hundreds and hundreds of gigabytes just because it's a six year old computer. And my God, it had. VS Code leaves logs like you wouldn't believe, I had 8 million simulators going back to iOS, whatever was out in 2017, so I got rid of all that stuff, um, it's, it's, it's nice to get that purge out.

That's great that you're down to four gig. I can't believe that the OS was probably actually, yeah, taken up 2x the space it needed to. That's funny. 

[00:27:40] James: It's intense. No, it's, it's fun to cleanse stuff overall and just get rid of. So it does make me want to like do a fresh, clean install sometimes. Like, oh, it's so fresh.

But yeah, I remember I was at work and my drive was. Because it was like drives between my personal account and my work, my work machine, it's like multiple visual studios and this and that. And then I remember like I had like the Outlook cache was like a hundred gigs. It was like every work email for like ever, every, you know, and every attachment, all this other stuff.

I'm like, Oh my God. You know what I mean? So just like wild. So, yeah, 

[00:28:15] Frank: but. Ah, fun. Aren't computers fun? Yeah, I just had, um, I was cleaning out a drive too on a, uh, you know how Linux machines never crash and they're super reliable and all that stuff? I actually had a hard drive failure, totally corrupted its own file system, and I've been trying to, I've been trying to nurse it back to health.

But in the meantime, I bought like another drive. And so I'm trying to like nurse one back to health just so I can get it running to transfer all the data over to the other drive. Um. It's, it's all just, I, I'm always surprised that like, yeah, the Apple device batteries die over time. Um, but I think this hard drive I had in that computer lasted a whopping two years.

And so I, I kind of almost get why companies like Google and Apple don't want to support hardware for, you know, overly long times. Cause At some point, things are going to start breaking. Um, but at least seven years of, uh, OS updates, assuming that your storage continues to work up to that seven year mark, hopefully you'll be able to get the update then.

[00:29:24] James: Yeah, I agree. All right. One of our last topics in case Frank loses the internet here. Um, I have been. And you've been playing around with a lot of smart devices recently. And I, I've, I've had, I have a whole series of things. So I have, I have some cues, I have some SW switch bots, uh, sitting around, uh, which are little devices.

I have their light bulb, I have some light bulbs. I have the little switcher thingy and I have the OPPE stuff, uh, which is like kind of smart thermostats and outlets and things like, and light switches and things like that. And I recently, Got really into some of the automation stuff. All right. Oh, I also have like a bunch of security cameras too.

So the smart things too. Um, got smart, smart sauna, for example. Uh, one thing that I started to realize is that it feels as though, I want to get your opinion, in this smart home, IOTified world that we live in, it feels as that all the ecosystems are like. 80 percent the way there, but none of them also integrate with anything at all.

Maybe I'm doing it wrong, but here's an example, Frank. Here's an example. So let's say I have some, uh, let's say I have some, uh, a smart light switch and I want to automate that to turn on my outdoor lights, let's say at dusk and turn them off two hours later. You can do that. That's pretty easy. It's, it's built into, into the apps.

It's there. Now let's say though, you want to add a motion sensor outside, right? Now you got to hope that this company makes a motion sensor that, that works outside as well, that you can then program to say, okay, after midnight, turn on the lights, if there's motion outside and then turn them off after no activity, after two minutes.

Now, additionally. Your house is probably surrounded by security cameras that have motion sensors in them. But these two worlds, Frank, can not talk together. They're not allowed to talk together. Right. No, they're not allowed. Now. I will say this. It's like, I am using, like, I got, I got HomeKit. on some things.

I got Google home at some other things. I got other, other ecosystems that are complete boxes that can't break down the walls on at all. Um, we got the YFIs. We got the ZigBees. We got the stuff, but nothing. I feel as though I got the things behind me on the wall that I'm like power, they have threads or Matt, they have some, I don't even know.

I even know what to integrate. Yeah. I feel as though. Everything's like 80 percent the way there. And then when you're like, Oh, this is a great scenario that IOT would be great for you, like run into a brick wall and then you don't know how to get out of it. And then what you're really thinking is, do we really need one mega company?

To just make every single device, right? Cause all these companies they're all in their own little ecosystems, but what if just one company made every single smart device humanly possible, that all works seamless and we could just go to one company. That's not going to exist, but like, do we need that? Am I bad?

Am I doing things wrong? Do I just give up at this point and say James, you know, put, put, put that indoor motion sensor outside because that's good enough. You know what I mean? Well, I'm 

[00:32:54] Frank: going to say a lot of words now, but none of them are going to help you. So. Good. Perfect. You decide what you want to do with this.

Okay. Um, yeah, I, I, I, I mean, we're all kidding ourselves when we call these smart homes because there's nothing smart about it. What we've done is we've had remote controls to all the light switches and lights and things in the house. We have a bunch of remote controls and I really do, um, think of it in terms of a kid playing with an RC car.

That's not a smart car just because you have a wireless joystick controlling it. All we did was upgrade the switches to our houses. You know, smart bulbs, all that, they're just switches. It's, it's, you talk to it over your iPhone now. It's just a switch. All we did was add wireless switches to all of our houses.

Uh, now is the hard part. How do you make that smart? And what is the definition of smart? So yeah, you've had companies since the beginning, like, um, if, if this, then that, you know, like to try to do this, but yeah, none of these companies talk to each other. None of them post their data. Yeah. So my solution has been to stick with the Amazon family of devices only because.

Then it's one app controlling the majority of the devices. And then you can add triggers and things like that. But yeah, gosh, darn it. If you want to use another brand or use some things from Apple or this company or that company, yeah, you really are stuck because what we have is a bunch of remote controls for our lights and none of the remote controls want to talk to each other and therefore you cannot create the, what you want.

is a little house brain, and a house brain is managing things, and it's making reasonable decisions on your behalf, knowing that you're a human being person, and that you're living in a house, and it should make appropriate decisions for you, like maybe turn the lights off at midnight if no one's moving around the house.

That would be a somewhat intelligent thing to do. Um, but we don't have that. We will. Um, and it'll cost 10, 000 and it'll come on a little Apple branding and it'll be like Siri 3. 0 or something like that. It'll come. We will get it. So maybe my advice to you is be patient and shut up. But, um, the truth is, it's just, it's sad.

Yeah. We've, we've rewired our houses, added a bunch of remote controls, but the remote controls don't talk to each other. Yeah. It won't, I don't think it'll be the dystopian future though. I think there will be a few companies that you can have to AI control your house and do all the smart, actual smart things 

[00:35:35] James: for it.

Yeah, I will still, I'll, I'll put this down. So I'll tell you what I'm doing. So I am, I still believe that HomeKit is. The best I love home kid, only because, I feel as though when you are, um, have a partner that is, um, Not into the same stuff. Like Heather could kind of care less about Smart Home stuff. She puts up with the fact that I wanted to do her Smart Home stuff.

Now there are some things that she likes like having us the, the, the Hughes in the bedroom that we can turn on. And turn off from the bed, right? That's really nice. Uh, before night boom tickle, like a thing, right? The bigger one. Right. That's a nice, that's a huge, nice touch. So I think that's really good.

Now that being said anything else besides that she could not care at all. You know what I mean? The thermostat's like, that's nice, but If I could just get a nicer thermostat, that'd be good too, but, but in general, she just wants a thermostat to work, you know? And, um, but HomeKit makes it so easy to share a home.

There's no other things to install. There's no account. There's not a whole thing. The UI is pleasant, you know, and integrates into the control center. It's just, it's a pretty delightful experience compared to, Oh, uh, just install these 18 apps and then bop around, you know what I mean? So. I don't know. I think that's really good.

That being said, it doesn't work with any ZigBee things. You know what I mean? So HomeKit can't like talk to any ZigBee things at all and integrate into things like a smart things does or a Google home or, uh, or the, the, uh, Echo family of things, right? So now I'm forced where I have all of these ZigBee things to get a Google home.

So I have, cause I just want something to talk. I don't want even, I guess I want to talk to it, but I want to be like, hey, hey, hey, Fit, did thingy go do my thing. I get jealous when you're like, Hey, thingy, go turn my lights to whatever. I Oh, that's nice. Oh that that's nice. But it's only nice when you have everything.

If you have nothing or one light bulb, it doesn't matter. But once you have like the whole upstairs there, that that's the key, you know, with dimmers and all this stuff, and they're smart. Now the other thing too, is. I, and I drew this diagram for Heather and I want you to confirm that I am correct. When I have my Sinope, um, hub, right?

My, my ZigBee gateway hub and all my Sinope products are talking to it, or Hughes things are talking to it. And I put that into the Echo ecosystem, or I put that into the Google Home or even HomeKit ecosystem. Well, they might, HomeKit might work different actually, but let's, let's go with Google and Echo.

Okay. They plug into the ZigBee, they plug into the Wi Fi, all that stuff. Now, is it correct that what's happening is I say, Hey, Dinghy, go turn off my lights. I'm then speaking to Google or Amazon, and then Google or Amazon is talking to Sonope or Phillips Hughes and then turning off the thing. Right? It's as if Google's talking to the app and then talking to the thinger.

Is that correct? Yes, yes, 

[00:38:50] Frank: and that path, yes, and that path can be short or long, and it can even have more hops in it than you're even guessing right now. Um, in the bad, bad old days, it would go, first your voice would go off to, let's say, Amazon. And then it would go off to another server, and then it would signal another thing, the hub.

The hub then would Z Wave, ZigBee, off to the actual device. Um, I myself, um, I do Z Wave and ZigBee from the Amazon device itself. So in that case, it doesn't have to do any of the middle hops, it can talk to the device directly. They're the more expensive ones, but some of them have those radios built into them.

But in your case, yeah, it's doing roughly what you just said. Pray that it's not doing the first one I just said. Hopefully what it's doing is, um, your voice up to Amazon, then dingus to the hub, hub to the device. But yeah, it's that many hops. And that's why, in general, I don't like 

[00:39:58] James: hubs. Yeah. Hubs are bad.

I got too many hubs sitting around. Now. I will say one other thing is. With HomeKit, it's different though, right? Apple is talking to your devices directly, 

[00:40:12] Frank: right? No, it doesn't mean any of that. Apple can still do all the same hops. Um, so a HomeKit device can be, um, uh, Wi Fi or ZigBee Z Wave kind of stuff.

The Wi Fi ones would talk directly to the device the same way my Wi Fi device or Wi Fi devices talk directly because it's Wi Fi. Everyone knows how to talk it. It's only when you're into the ZigBee Z Wave where it's a special radio. So, for example, the Apple TV may not have the special radio needed for ZigBee Z Wave, whereas the um, What's the, what's it called?

The home pod, whatever the, the, the HomeKit Banger Home Pod. Yeah. Yeah. That, that thing might have the radios. I honestly don't even know. Um, so it really just comes down to the actual physical wireless communication that you're using. Again, another reason why I went wifi, because I didn't wanna deal with any of 

[00:41:05] James: that

Interesting. That makes sense. Ugh. Oh my goodness. All right. Well, that I think is going to wrap up this, uh, merch conflict on that delightful note, if you probably have some, you know, Insight into any of the topics we talked about, especially the smart home stuff. We've done an entire podcast on smart homes.

And now that this podcast, like 10 years old, like, uh, I can't even imagine going back and listening to those, right. Keep doing it. Well, cause it, you know, it changes and evolves. You know, I have a bunch of devices behind me and it's also more coming in the mail that I'm excited to install. Uh, which is all the Zigbee.

So I don't have like a thousand. Well, I think there's a maximum that the hub can actually talk to. I think it's like. 32 or 64 or something like that. So I figure what the maximum is. And then you need another gateway, I guess, after that. 

[00:41:51] Frank: Yeah. Yeah. And I just redid all my stuff because I changed my mind how I wanted to do any of it.

So yeah. Time's a change. 

[00:42:00] James: You're done with wifi. You're on the ZigBee train, Z Wave, ZigBee train. What's going on? 

[00:42:04] Frank: No, no. I used to go smart light bulb, but now I'm smart light bulb controller, dumb bulb. I'm, I'm on the dumb bulb bandwagon these 

[00:42:14] James: days. I agree. The, the, the dumb bulb, I think is the way to go because there's no difference in Heather going to her phone and saying, turn off the, turn off the thing, the light bulb, um, at night versus turn off the switch.

There's no difference there, right? The only difference is that no matter what light bulb you put in there and change your fuse, do whatever, it's like. The light bulbs are then cheaper and you can kind of customize it. And I think there's better in general, you would try to convince me that the do make, make every outlet in your house.

I was redoing all the outlets in the house, redo every outlet in the house as a smart outlet, which I did not do. Thank goodness. Cause it would've cost me a billion dollars. But that being said again, I have now a bunch of. Uh, smart outlets in there. So the outlets that you plug into the outlet, which is cool enough.

And I'm going to put that into the places that I think are really important. Because, turn it off at the source, you know what I mean? Um, as Mike said. 

[00:43:16] Frank: Yeah. Um, I, I forgot in a house just how many light fixtures can be like a four bulb fixture or a six bulb fixture. And so, yeah, putting a smart bulb in each one is just a bit too costly.

And honestly, I think it's just more elegant, uh, to control it at the source. As you said. 

[00:43:35] James: And my new favorite thing, by the way, is that, I mean, you almost want every single thing in your house to be this way, but it's really cool when the devices can give you additional data that you couldn't get from a.

Non smart device. So for example, in the Sinope app, uh, they actually have a protocol with their devices that tell them the energy consumption for all of the devices. And then you can say how much your kilowatt hour is, and it'll tell you, um, pretty much like how much. How much it is. And then they also put onto the same graph for like temperature outside or whatever, so you can say, okay, versus the temperature outside or using more heat, how much is that costing you?

And you can do like a two year diagram basically, which is really cool. And it'll show you every single device, how much kilowatt usage it used, and then how much it costs you, which is just pretty cool. Nice. 

[00:44:35] Frank: Wave of the future. Yep. Um, it's kind of, I want that feature. I don't think I'll get that feature.

That's going to take a lot of work for me to have that feature. So that's not going to happen. But, um, I think we should do a future episode of Merge Conflict. We should find out if we can access some APIs or something like that, and maybe add some smarts to our Remote controlled homes, since we've installed all these remote controls, you and I should see about if we can actually, uh, program in them and automate them and make them do interesting things.

[00:45:05] James: Yeah. You know, what is wild is, um, I was talking to the SwitchBot folks. Now the SwitchBot folks make all sorts of things, but the original one is a little switch that turns on and off. And I, that's what I have. And it adds to some of their like smart light bulbs, but. They have a completely open source, um, API, uh, as well.

I mean, I don't know if I can find it, but they have a NET library. Um, switch bot. net API, GitHub, if I can find it, but basically it has all this stuff built in. Oh, here it is. Yeah. Wonder labs, uh, switch bot API, and they have basically everything inside of you. I'll send you a private chat here so you can see it, but authentication, every single device, everything's API forward, XYZ.

And they have, yeah, C sharp examples, Python examples, all this different stuff. It's pretty wild, but I didn't do anything with it, but I was like, yeah, that's pretty cool. That's pretty neat, pretty neat. So I like that. Well, 

[00:46:05] Frank: we'll, we'll, we'll turn you smart. We'll turn your house smart somehow. Drag 

[00:46:10] James: it into smartness.

All right. Well, right into the show, mergeconflict. fm. You can become a patron subscriber. You get bonus exclusive episodes every single week that we put out here. Um. If they work on Frank's sweet, sweet Starlink internet. Uh, you can also tweet at us, uh, post, I guess you can post at us on the Twitter X and you can, uh, follow us and get, uh, videos on YouTube.

Uh, we have a YouTube, which is youtube. com forward slash at merge conflict FM. That is where you can find us. You can. See our faces as we talk. Um, and let us know what you think of this format. We're recording this one in StreamYard versus our Zencaster ones here and there, let us know. Um, and that is the point to do it for this week.

So until next time, I'm James Montemagno. 

[00:46:52] Frank: And I'm Frank Krueger. Thanks for watching and listening. 

[00:46:56] James: Peace.