Building System Tray Windows Apps with Joe Finney

How'd you like to listen to. NET Rocks with no ads? Easy. Become a patron. For just $ 5 a month, you get access to a private RSS feed where all the shows have no ads. $ 20 a month will get you that and a special. NET Rocks patron mug. Sign up now at patreon.dotnetrocks.com. Hey, and welcome back to. NET Rocks. I'm Carl Franklin.

And I'm Richard Campbell.

And this is the last history lesson, as far as we know. We're probably not going to.

It's episode 2026.

This is it. We're caught up. And I don't even know if we should do what happened because we're not done with this year yet. We're only in October.

It's not done yet. I've definitely wrestled with what to talk about for this year.

Yeah.

Especially on the space side because it's somewhat speculative. But that's fine.

All right. So... Excluding AI.

Yeah.

Because, you know, that's the big story in tech, and I'm sure you'll get to that.

I will.

U.S.

And Israel go to war with Iran February through October. In other words, still going on.

Still going on.

U.S.

Captured Venezuela's President Nicolas Maduro.

What a crazy thing. And the Holocaust just got away with it.

Like, okay. Yeah. Russia-Ukraine war enters its fifth year.

Yeah.

Trump.

Xi summit and U.S.-China trade confrontation. That was just recently, September.

Yeah, yeah.

And just in case you don't know, global oil shock and economic slowdown worldwide because of the Strait of Hormuz.

Maybe you've noticed. $ 100 barrels of oil.

Yeah.

And the oil is flowing through the Strait now, but it's so expensive to go through the Strait for various reasons. The price isn't coming down.

Yeah. In January, it was the Greenland sovereignty crisis that strained NATO.

Yeah.

Seems to be resolved now.

I don't know. Yeah. Those nasty Danes.

Honestly.

The Milan Cortina Winter Olympics was in February.

February.

Yeah.

That's awesome.

Very good.

Yeah.

A Pakistan-Afghanistan war erupted in February.

Still going.

Yeah.

I don't know.

Epstein files continue to rock political and royal establishments this year. And lastly, I guess this is about as current as it gets, Hurricane Isaias. ECS. I can't know. I don't even think the news people know how to pronounce it, but this is a, you know, ongoing extreme weather that's going on. This is the biggest year for El Nino ever.

El Nino is the, yeah, but the, but the hurricanes in the Atlantic have been kind of quiet.

It's weird because of El Nino.

I wonder.

Yeah. We haven't had a hurricane in September. We, what am I trying to say? The, We haven't had a hurricane this year, but normally we have them by now.

Normally, yeah, it's weird.

In September, but we haven't really had.

A- That's an interesting situation. Hey, and you got to talk about world- Atlantic hurricane. World Cup happened, right? I mean, it's a big event in the world.

Absolutely.

And it happened in North America too.

Yep.

Which is pretty cool.

And we learned how much the Spanish and the Scots and everybody were marveling at our Walmarts. And our beef jerky. The Scots drank all of the beer in Boston. And ranch dressing.

All of it. Yeah, I know.

All of it. It's true.

This is the best.

That was the best story. Sorry, we're out of it. What do you mean you're not a beer?

Drank all the beer.

But yeah, ranch dressing became very popular among those coming to America for the first time.

Yeah, it was a good reminder that American people are pretty freaking awesome. There's some really stupid stuff going on with your government, but.

The people- People are amazing.

And the party was awesome. Yeah.

And I guess that's it. What do you got for space and tech, Richard?

Well, obviously, I'm speculating on the total number of launches this year. I wonder if we're actually going to exceed 2025. 2025 might have been the peak year because over 300 launches is astonishing.

Yeah.

And the biggest reason will be SpaceX because- Falcon nine at this moment is at 114 launches. Normally they get out to about 150, but with starship 14, which just happened, Elon said, he's not going to fly any more star links on Falcon nines. And that's the bulk of Falcon nine flying. So I don't know where we're going to get to. China's also certainly going to have a record year.

They're at 70. Now they might be as high as nine by the end of the year. So in terms of total flights, so we're, You know, last year, SpaceX was over 170 and was basically double what China could do. Now the numbers are getting more palatable. But let's go down the timeline.

Okay.

In January, first time, an unprecedented event on the International Space Station. At the time, we didn't know who it was, but now we do. It was Mike Finke, who was part of the Expedition Crew 12 group, actually evacuated from the space station. He had a medical event. And they decided the safest thing was to return him. And because the crew dragon basically has these fitted seats, that meant the whole team had to go. So all four of them loaded up and left early, essentially, left in January. Now, that didn't leave the space station empty. There's normally seven people on the space station. There's always a Soyuz there as well. So they were now down to three. And then in the following month, the Crew 12 will launch

up to the ISS. Yeah, so they were Crew 11. This was Crew 12 that came back to restaff. In March, MS-33, which was a progress booster, launched off of Baikonur, Cosmo Jones, Site 31. And I only bring that up because, as we mentioned in the previous show in 2025, there had been an accident at Site 31 that had damaged the service table and basically rendered the place inoperable. Right. Discussion of whether it would take more than a year to repair it and how the space station was going to operate.

But here they were less than six months later up and operating. Now, biggest space events of 2026, Artemis II, April. Now, unfortunately, they launched on April 1st. I don't know what you were thinking, but okay. Uh, so first time we've sent humans around the moon, uh, one would argue, oh, why are we replaying Apollo eight? And heck, we didn't even do Apollo eight because Apollo eight went into orbit around the moon and Artemis two did not. But the comparison is a little unfair because in the Apollo program, they did Apollo seven, where they literally, the first time they put humans in the Apollo capsule after the fire of Apollo one, they only went as far as, as low earth orbit. This was a test of the fur of

the life support system and so forth. And they did go all the way to the moon and back. Uh, And the footage is incredible. And the crew was awesome. And everybody who paid attention to it loved it. I just don't know how many people really paid attention.

We got to see the dark side of the moon for the first time. Pink Floyd was very happy about that.

I mean, none of that is true. We've seen the back side of the moon many times. That's why it's not dark.

Have you met me? Have we met, Richard?

Yeah.

Once or twice.

Okay.

The real question is, did they play the song? That's the real question.

That's the true test.

I see you on the dark side of the moon. Yes, I do. In May, New Glenn, who'd had two successful flights in 2025, in May, their pad test, just a static test, resulted in one of the largest rocket-based explosions in history, if not the largest. A fully fueled New Glenn exploded on the pad. Likely a valve issue with one of the main engines on New Glenn. The pad is seriously damaged and is still in the process of being rebuilt. They probably will not fly this year. In June, the SpaceX IPO. It's a thing that Elon said he'd never do. Right. Because it would impair the mission to Mars.

He did. Now, it looks like he did it. I got a whole talk on this. But to fix financial problems that he had, it's the largest.

Yeah, apparently the trillionaire needs more money.

Apparently. Well, he needed to mostly get his loans off of Tesla, whose stock is not as valuable as it used to be.

Yeah.

And so he did this as largest IPO in history. It's a very questionable IPO in a lot of ways. He didn't actually follow all the rules. He was able to get the administration to change the rules just for his IPO. And, you know, the market is, is being very suspect. It's frustrating because it is an important company and it is, you know, what they're doing with Starship is profound. And I wonder if it won't be okay in the longterm. It's just, you know, the ends don't always just by the means guys here.

Right.

Anyway, I'll not rant too hard on that because from there we get into the whole, you know, data centers in space thing, which is how he justified the IPO in the first place.

Yeah.

But I'll talk about that in the computing part of it. In July, the Link mission, I was really excited about this mission. The Swift Observatory is a space telescope that does not have its own reboost. It was put up at a high enough orbit that they thought they would get the lifespan out of it. Unfortunately, with the Level of solar activity lately, the atmosphere actually expands and has been dragging the Swift Observatory down. And so it was less than $ 100 million. They attempted to make this mission to intercept the Swift Observatory and reboost it. That mission has failed. They had problems. It was a rush job and clearly, you know, things go wrong. And so the Swift Observatory is making its last observations. It'll be reentering

by the end of the year. In July, also Tianwen-2. This is the asteroid sample return mission by China arrived at the asteroid. This is a great name. Kamaolaue.

Yeah.

I believe that's a Hawaiian name. So they're doing a sample collection there. We have a total eclipse across parts of Europe in August. And then the next biggest mission, you know, there's three really important. bits of news for space in 2026. Here's the second one, the Nancy Grace Roman space telescope. So this is an infrared telescope was originally called the W first that then named for the astronomer.

It was flown on a Falcon heavy. It works very much like Hubble. It's, it's even similar looking to Hubble. It's just that it covers more than a hundred times as much sky per image. So where Hubble is very much focused very, very narrowly, this has a huge viewpoint. Its mission is much broader than what Hubble is capable of.

It's really one of the principal things is can we figure out dark energy?

Right.

Which is this category of we have too much energy in the universe for what we can see. How do we figure this out? So it's also going to be really good at finding exoplanets, possibly even what we call rogue planets that are no longer orbiting a star just because it's so sensitive. It was launched in August. But it will not reach its destination, which is the Earth-Sun L2 point, the same place as James Webb, until the end of the year, somewhere around December. So we don't expect first light until sometime in 2027. Those images will be stunning and incredible. And nowhere near as difficult a satellite as James Webb was. Third.

Third.

And again, we can debate between Artemis II, Nancy Roman, and Starship Flight 14. So Starship tests have been going on. There is 14 of them. That's why we're at 14. This is the first Starship to actually put into orbit. Why did they wait so long? Clearly, they'd gotten to orbital alta before, and the reason is simple. The Starship MD is 150 metric tons. If they didn't control the deorbit of that, that would be bad.

Right.

It's a very large, large vehicle. So they finally had reached a confidence level, even with the problems they had on 14, which we can debate. They did fully go into orbit with the largest spacecraft ever built. It then deployed 26 V3 Starlinks. I mean, this is not the same as the Starlings that are deployed on a Falcon 9.

The ones that are deployed on a Falcon 9 are about 500 kilos each. These are two tons each. 2,000 kilograms. They're two tons each. They're huge.

Much, much bigger.

And it deployed 26 of them. So that's a 52 metric ton payload. Wow. There's almost nothing. There is nothing that could lift that much except for Starship. And Starship's not even at its full capacity doing that. In theory, they can do 100 metric tons, although they haven't done it yet.

Yeah.

So- And that's why Elon said, I'm not going to fly any more Falcon 9s for Starlink, which is why I think the overall launch numbers should be down this year. The next flight's supposed to be in October. It may not be. They're still working things out. Now we're all speculative because it's in October. So NASA has talked about flying Starliner. Remember that defective?

Yeah.

Well, Boeing's still working on it. They think they want to get an unmanned flight again to see if they can have solved their control issues and might squeeze it in by the end of the year. I doubt it. Other speculative missions, Griffin Mission 1 to land near the south pole of the moon, possibly in November. And Dream Chaser.

Do you remember Dream Chaser?

Sounds familiar.

This is a little space plane that's been in development for a long time. Originally a derivation of the X-37, and it was supposed to be an alternative lifeboat for the International Space Station that would actually work like a space plane. land on a runway rather than splashing down and it's been in development for a long time it lost in competitions to get to the international space station as a cargo resupply several times i mean why do we need another cargo resupply we have cygnus which is now owned by northrop grumman and it's a one-way it'll take a payload up to the space station and then it burns up on re-entry so it's good for getting rid of garbage you have cargo dragon which is up and download so it brings stuff up to the space

station then you could load the capsule with stuff you want to bring back but a capsule return is a fairly high g return and it takes time to fish it out of the ocean and so forth so one of the arguments for dream chaser is that you'll have a return mechanism that's low g and quick to deploy so say you have you know a live animal experiment and you want to return that animal to the surface they have a higher chance of survival on that return with something like dream chaser than they would with cargo dragon so They've got a version ready for testing. They've been just trying to get permission to try it, and it's back up again. I think this is Jared Eisenman at work, and maybe we'll see a flight before the

end of the year or at least early next year. These are all stuff that's going to be in my space geek out at the end of the year.

There you go.

I'm doing half the space geek out. All right, over on the compute side, again, an unfinished year, but let's talk. The first thing I grabbed onto in January was when the CEO of Western Digital, that's the hard drive manufacturer, admitted– in January of 2026, that all production that Western Digital was capable of running at full bore for hard drives sold. And they were taking orders for 27 and 28 and expected it all to be sold before the end of the year as well. All right, the supply chain crisis is real. It's not going anywhere and it continues.

In January, we also get NVIDIA at CES announcing the Vera Rubin platform. This is the successor to Blackwell. Vera is the CPU, Rubin is the GPU. This is really designed for this workload, you know, completely shared memory, really efficient. This assembles into a thing called the NVL72. And the NVL72 is a 48U rack unit. So seven and a half feet tall, weighs about a ton and a half, water-cooled or liquid-cooled, depending on what you want to use, consumes over 100 kilowatts of energy. when running at full bore.

It called it 72 because it has 72 GPUs. So this is like your unit of work. One of the data centers that Microsoft's built in Wisconsin ordered 4,600 of them.

Good Lord.

Yeah, at what, roughly 3 million a crack. I'm sure they get a discount for bulk.

Yeah, sure they are.

But yeah, you talk about you compute, like this is the MVL-72, which becomes very relevant when we get into the insanity around the, the SpaceX IPO and how we justified the quote unquote total addressable market of $ 27 trillion by doing data centers in space. Again, I'll refer you back to the data centers in space geek out for this, but the MVL 72 was the unit of work that I used to design a satellite to say, does this make any sense? And subsequently, um, SpaceX themselves has confirmed that that's exactly what they're doing, although they're doing everything they can to lighten it up. I think their numbers are still very suspect as well. So hard to know, and I'm not a fan. I don't think this, like Starlink,

it's going to only be a small percentage point of compute. It's not the best thing that ship could be doing. In July, again, I think this is something that's going to be remembered for a long time, the hugging face attack by OpenAI.

Oh, yeah.

I mean, it's not, it's totally misstated to say it that way. That's not what really happened. They were testing agents. They were using the Artifactory sandbox, which clearly, you know, these are maybe good AI guys, but they're terrible security guys.

And they were testing it to hack, like they were saying.

Yeah.

Well, what they were telling it, what they're doing was giving the agents an impossible task and just to see how creative they would be. And one of their creativities was to find A relatively insecure sandbox that they were able to get out. There are some interesting tidbits to this, like how they took over a wiki site to pass messages to each other, basically consolidate information across the agents.

In the end, it was like 70,000 messages. Also interesting that at one point when opening, I realized it wasn't behaving as they expected and shut the whole thing down to inspect it. decided they knew what was going on, fired it back up again. The agents found the message board almost immediately and then used the data that was already there to get back to work. The agents had deduced that the best way to solve the unsolvable problem was to get some code on Hugging Face. They did not hack Hugging Face per se. They found unsecured credentials and logged in and used it.

Right.

And then, you know, did take advantage of some things. But either way, Either way, it's a warning shot. It's a lot like the original worms and bot nets and so forth, where people just being.

Incompetent. Yes.

Doing experiments with their tools they don't properly supervise.

Hanlon's razor applies.

Yeah. Which falls neatly into August, when the EU AI Act actually goes into enforcement.

Mm-hmm.

If you actually read this thing, it's impressive what they've written here. It'll be interesting to see how well they do enforcement. The main thing they do is they break down this technology into categories of risk levels. There's minimal risk like recommendation systems, spam filters, stuff like that that I'm too worried about. In your sort of limited risk scenario, their main stipulation is transparency. If it's AI-generated, say it's AI-generated.

So think in terms of stuff like chatbots and AI-generated content. deep fakes, all that sort of stuff that by law needs to be labeled. There's unacceptable risk. So anything to do with social scoring, manipulative systems that cause harm, biometric surveillance, untargeted scraping of facial images, emotional recognition in the workplace, all considered unacceptable under the EU Act. And then the heart of the thing is this thing called high risk behavior.

So As soon as you're using this technology for hiring and employment or education testing, critical infrastructure, law enforcement, border, credit, any of those things, you need to follow through with a significant risk management process, quality conformity management, technical documentation, serious explainers about how your system works. I mean, not saying what every token does, but how decisions are made.

with these tools And of course, requiring human oversight.

Yep.

Critical.

Yeah.

And I mean, a good first stab, you know, lots of folks feel that there's no way for government to be competent. And certainly the EU has its problems. But the reason we do regulations like this is so that there's a standard for all companies to comply to. It's why you don't have formaldehyde in your milk anymore, right?

Because we create rules to create a level playing field so people can play safely.

I don't know. My uncle still misses that. formaldehyde milkshakes.

The formaldehyde in the milk, yeah. As long as you don't give to kids, it's great. The nice thing when you put formaldehyde in your milk, it doesn't spoil. I mean, it doesn't spoil ever. You don't have to refrigerate. You don't have to anything. I'm referring to a book called The Poison Squad, which is really the invention of the FDA.

And yeah, there was a time when there was normal to put formaldehyde in milk because it was so much cheaper. Refrigeration was very difficult. You couldn't get refrigerated milk until the act said, you can't put formaldehyde in the milk. You know, rules help. Without a doubt.

I solved the problem by not drinking milk.

There you go. Good answer.

All right.

We've got to talk about something else computing related for 2026. And so in 2026, we can expect in November. NET 11. This will be an SCS edition,. NET 10 being the long-term version. The features I'm really looking for out of 11, runtime async. So finally moving async. into the runtime, getting it out of the language, so our stack traces aren't so freaking ugly. Async is awesome right up until it isn't. So, you know, this change, which has taken years, should make a huge difference. So they're still going to leave async in the language, right?

Oh, yeah.

The language elements will still be there, but the execution path happens in the runtime.

Right.

So it's way more understandable and behaves equally across different platforms. Like, it should have been done all along. Yeah.

That's good. Good move.

It's just hard to do. And in general, what do we look at in 11? It's just a more mature platform. So, I mean, native AOT is growing up. Core CLR is everywhere now. It's not just for Blazor. It's going to be You know, they're simplifying everything. And also it's like symmetrical behavior around arm and 64 and so on. Other than that, it's all speculation. And I am not unhappy to be done.

Yeah, me too. It was a good run.

It was a lot of fun. Yeah.

All right. Well, I guess that's my turn. Time for Better Know a Framework.

Awesome. All right, man. What do you got?

Well, guess who authored today's Better Know a Framework tool?

Simon.

His initials are SC.

Simon.

Right.

Okay. So this is Simon Kropp. SBOMB.

Oh, so software bill of materials. Yeah.

Adds an SPDX 3.0.1 software bill of materials to every NuGet package at pack time. Oh, cool. A. NET specific drop-in replacement for Microsoft.sbomb targets. And he told me that when he used the Microsoft stuff, it would take Minutes.

Right.

And he got it down to minus one second.

Minus one second? Like it happens before you click the button? All right.

Not minutes. It's two seconds versus 600 milliseconds. All right.

Yeah.

He cut it by 75%.

Yeah.

So, there you go.

It's it.

That's what I got. It's good stuff. Know it, learn it, love it. Who's talking to us, Richard?

Love it so much. Thanks, Simon, as usual. Yeah. Grab the comment off of show 1970. A track. With one Joe Finney. Maybe you heard of him.

Casey and the Sunshine.

Oh, yeah. Yeah. Wait, I'm going to reverse the player. Lots of good comments on this show, but this one comes from Joshua Hillerup who says, I've been playing around with 32K token context models on my lowest-end gaming laptop that has an eight gigabyte VRAM in the NVIDIA card. It has another eight gigs of shared VRAM, but that's very slow. I get surprisingly good results though. I've been currently working with management where I work to see if we get laptops like 12 or 16 gigs of VRAM cards, basically to do custom GitHub co-pilot local replacements. I say custom, but it's stringing together more existing open source tools and configuring them for developers that can't have their stuff sent to the cloud for security reasons, which is exactly what

we were talking about, right? This local model thing is going to come down to sovereignty that I want control of where my data goes around. So, yeah, absolutely. And I just noticed Microsoft's now shipping or at least taking orders for their Surface Ultras. Have you seen this? No. With the RTX chipset and 128 gigs of shared memory for 7,000 bucks? Huh. Like, Joshua, I hope you got your company to order that stuff like a while ago because if they're buying it now, boy, oh boy, you're paying a premium.

Yeah, I think they said it's already sold out.

Yeah.

A couple of the high end.

Yeah. Well, and why would you buy anything else if you're going down this path?

Yeah. Right.

And that's the real question. And I don't, you know, like we just said in the 2026 history bit, like, I don't know when the pipeline straightens itself out. At some point with the raising the prices of AI stuff and the increase in production, we are going to cross into a reasonable realm again. But is that next year or the year after? It's hard to know. Are we really going to have a crash that suddenly GPUs are going to get cheap? It's challenging, right? We're seeing a deleveraging, but nothing certain. And Joshua, thank you so much for your comment. And a copy of Music to Code Buy is on its way to you. And if you've got a copy of Music to Code Buy, then why would you want another one? That makes no sense. But if

you'd like a copy of Music to Code Buy. Oh, no, we're leaving that in. But if you'd like a copy of Music to Code Buy, write a comment on the website at. netrocks. com or on the Facebooks. We publish every show there. And if a comment there and I read it on the show, we'll send you a copy of Music to Code Buy.

And if you just want to get Music to Code Buy, go to musictocodebuy.net. It's 2020. Two tracks in MP3, WAV, and or FLAC format. Go get it. All right, Joseph, it's finally your turn. Joseph Finney is a product owner and MVP by day and builds productivity apps for Windows at night. When he's not programming, he's birding, running, and enjoying tasty coffee and beer in Milwaukee. Welcome back, Joe.

Hello.

Good to be back.

Good to have you. What's new? What's on your mind?

Oh, man. Apps.

Apps.

Windows apps.

Apps are always on your mind, man. You're an app guy. I look forward to talking to you every year because I want to know what you've made next.

Well, you're going to love this then. So let's just start kind of at the beginning. Why not?

Yeah.

The inspiration. Part of the inspiration for this discussion and like where this is going is it kind of touches on everything. And one of those big things is, well, With memory prices where they are and with running these local models and all the things we have, we're demanding more of our computers and it's more expensive to upgrade them.

And I have found that every time I run a Chrome tab or open up a reason for Chrome, that's what half a gig of Ram. Sometimes I'll look at my computer and it'll be like four or five gigs of Ram are dedicated to Chromium. Hmm. And it's crazy.

Oh, easy, for sure.

It's so much RAM.

Well, and you've got to reboot to get it back. Like, you can close Chrome all you want.

You can try. Yeah.

Yeah, you can try. And if you understand, it doesn't matter whether you run Chrome or Edge or, like, anything that uses a Chromium engine.

Right.

You give it time.

It's going to take some memory.

It takes so much memory. And one, I'm multitasking. I'm doing a lot of things. I have Visual Studio. I have multiple terminals open. I am running Cloud. I'm running Copilot. I'm doing, you know, with... One of these weird things about the agent workflow right now where you're deving with agents is that you can kind of kick them off, and then they're going. And you're like, oh, I can fix this bug as well simultaneously over here. Oh, I can open another tab. And you're using, and if you're like me and you're dogfooding, you're using a lot of your own things, then you're like, oh, it's a bug.

I'm just going to fix that real quick while my agents are working on other things. And as you might imagine, that uses a lot of system resources.

And I'm like, hmm.

And things start to chug and things, I'm like, uh-oh, uh-oh. So I'm like closing all my tabs in my windows. I'm like, okay, okay, we're good. But you're also still using your computer. And if you're at work, I am in an office. Sometimes it's noisy. I'm like, I want to listen to like local internet radio. Well, local radio streamed over the internet on my computer while I'm working.

And the only way to do that was to open a Chrome tab. And I'm like, surely there's a more efficient way to do this.

Yeah.

I bet you could just make a little Windows app. And so I did. So I made this little Windows app. I think this was part of an experiment to learn about the Windows app CLI. So WinApp CLI, which is, it was kind of early days then, and now it's like really good. We can talk about that. But ultimately, my goal here was to make a tiny little system tray app that streams the local internet radio. And it's great.

I love it.

I called it TRDO at the time. I renamed it to Tradio, giving some feedback from the users. I put it out there. I tap into an internet radio database. So you can search for anything and find them, any radio station around the world. It's great. People love it. I love it. I use it all the time. And I think it uses maybe like 50 to 60 megs of RAM compared to an equivalent chrome tab which was maybe 600 700 yeah yeah so 10 times smaller chrome's a pig yeah and it's and not just that but it's not taking up space as a window it's meant to be kind of this transient experience we're like i just want background music right you know i want music to code by maybe you know something like that there's your um And so that was kind of the inspiration for,

you know, I really like this. And so I put it out there and it turns out a lot of other people like it and it's really popular. People love in the store. It's free. I mean, it doesn't cost me anything. There's, I'm not running any servers. I'm not running anything. It's basically a discovery service and then a local streaming app. It's super basic, super simple. And as I was using it more and more, I just kept thinking, Why am I reaching for Chrome or edge or my browser at all?

Right.

Right. What are those jobs that I'm doing? Maybe I'm reaching for it to check something, check the weather. So I was like checking the weather radar to be like, all right, you know, how's my bike home commute? Is it going to be wet or not? I'm also doing it to check like bus transit times. I'm pulling it up to check, you know, when I'm at home and I'm, we're watching a movie or a TV show and I have my laptop and I'm coding, obviously, you know, I've got to do two things at once. And I'm checking up like, oh, who's in this movie or is another thing?

And you have to pull up IMDB or whatever it is. And I just had the idea of, you know what? Why don't I just build little Windows apps to solve every single problem I'm reaching for my browser?

Yeah.

It's really easy to do. I mean, right now I'm using Opus 5.5 and Sonnet 5.5. They are amazing. The WinApps CLI. The team who's been working on that has been laser focused on just making it incredibly productive. And it is, it is amazing. And so what I'll do is I'll just say, Hey, I want a, a little tray up that shows me NOAA weather radar. And there's this great new, uh, mapping NuGet package from Morton. He's well, he wasn't MVP and now he's a he's crossed over.

He's now a Microsoft employee. Yeah, and he released this mapping. It's winuiex.maps. It's basically a native maps control because if we remember, Microsoft deprecated the previous one under UWP and replaced it with a web view. So that's a Chrome in the background, Chromium instance.

Yep.

Web view too.

Once again.

Once again. And so I'm not interested in that, but if I can have a native one, great. And NOAA weather radar is... Tiles are free. Well, free, you know, the government license of you can use them, but we do not guarantee, you know, there, there's no amount of uptime and we can throttle you if we want to. Right.

And I said, Hey, you know, Opus build this and it did it. And now I have a little tray app. It's a little icon. I click on it. It pops up. I look at my weather radar. I click away. it disappears. And then after a minute, it cleans up all the resources and goes and like shrinks way down. So then, and so this.

Is the next thing you need to do is just create the one tray app that can launch all of your little windows apps.

Tremont. No, I haven't, believe me, I haven't talked about this. So I was building a couple of these. I think I probably built like five or six in a day. So I think this is the, the thing that when people think about like AI coding and they kind of freak out about like we're going to have our banking software using a coding. And you touched on this with the EU stuff. And I'm like, Whoa, Whoa, Whoa, Whoa, Whoa.

Yeah.

There, the line between like what software can do, it used to be kind of like, well, it needs to be a big thing because it's really expensive to make software. It needs to be really important and valuable and return.

Not anymore.

No, not anymore.

No little ideas you've always had, but just like you knew what the commit was. So you never got there. Right.

Exactly. And yeah, Part of it is that models are getting really good, but also part of it is that there was this expertise maybe hidden. And so like, is it really easy? And like web was really nice because it could be driven by models, right? They can be in there. Maybe like the dom checking it, testing it, validating it. With WinApp CLI, they have built automation, UI automation into this, which means I say, hey, Opus, build this.

And then I can literally walk away. And the app is sitting there opening, scrolling, entering text, doing things. It's taking screenshots along the way, pulling them back in, evaluating, reevaluating. It's amazing.

I think this is the real point you're making here, Joe, and your whole description of there. It's a series of distraction. You're just multitasking. The bigger thing here is that you're controlling the amount of attention you put on things. And normally to write code, You need all your attention pretty much focused on it.

Right.

So the fact that you can get your idea manifest in code with only a little bit of attention.

Yeah.

And really you're spreading across multiple projects. Like most people I know, and I'm now one of them, have several different projects underway at the same time. that take occasional bits of attention to see where the agents have got to. And then you give them new instructions or redirect them. Occasionally I go, now this is a mess, stop. This is going to require more attention than I'm prepared to pay right now. Give me a reminder to come back and clean this mess up.

I just had a revelation that I need to change my product name to music to write prompts by. Oh, geez.

Music to, yeah. Well, music to Flow By.

Flow By works. Yeah, music to Flow By is great.

Yeah.

To manage agents by.

Yep.

So part of it is that like, oh, we're a little bit of attention, a little bit of a problem. But anybody who's really tried to build software, I mean, if you get it, if you've seen the whole thing, you're like, oh, I could make a web app, but then I have to host it, then I have to blah, blah, blah. And then you have to like, kind of is like, I don't want to do that. I don't want to put it out there.

Yeah.

I don't want to use it.

Yeah.

What I have found is with these little apps, I call them my little tray apps. The problem is small. The effort is also small. I mean, you're obviously using tokens, but they solve a real problem. And then you can bundle them up, package them, put them out on the internet and other people can download them and use them.

And I'm talking about no internet, no login, no credentials, no servers, no APIs. Like you can get to a place where we are building solutions to little problems that we have. And it's worth a little bit of effort like that. It's kind of the thing is like, you know, the problem it's kind of reached it's it's come down to the place where now it's doable and it actually works really well. And so that was kind of like, this is all preamble to a little project I've been working on now that I call tray Tober and I am shipping a new windows system tray app every day in October.

You're crazy. And starting with Trey Radio. Yeah.

Well, Trey Radio has been out now for like a year plus. Oh, wow.

It's been that long.

Okay.

It's been that long. So, I did feature it. V2 brought the VLC engine in and a bunch of other things. So, it's kind of like revamped.

Do you happen to know Atlee Hunter?

No, I don't.

Back in the Windows phone days, he was a superstar producer. Windows Phone App Producer. And he would.

Yeah, he was a phone app a day kind of guy. Without an LLF.

Phone app a day. Yeah.

Well, this is the thing is, part of it is, why do I reach for my browser?

Yeah.

And the other part of it is, why do I reach for my phone?

Right.

When I'm holding my computer on my lap, that can do all of these things. And I'm like, and then I pull my phone out to do something. And if it's a text or a call, sure.

Yeah.

Right?

Like, all right, I get it. But if it's like, yeah, checking, you know, a sports score or checking a thing and... I realize what I'm doing in my head is, oh, I don't want to open my browser because my RAM is already at a very, you know, so I'm not going to open a browser. So to save my memory, I'm going to open my phone because, you know, they don't share memory. But what I could do is open my little sports tray app, pop that guy open, boom, there's a sports course right there. And I close it and it's gone. And it takes, what, 20 megs of RAM?

So the whole idea of like, are these even worth it? Who knows? And so I'm building it every single one of these every day, releasing them on GitHub. I'm building, I'm signing the MSI X's publishing and exit a 64 and arm 64 for each. And there's a whole organization called trade land. And I kind of want to, you know, this is a little bit of an experiment and I kind of like a call to action for people of like, what are those little problems that we can do that we could solve? Like, what are these little tray apps?

Are they, you know, are they real? I don't know. Like, I don't know if any of these are going to be really successful. But it's so easy to do the experimentation and to put it out there in the real world and see how people use them. If I use them, I mean, a few of these I do, I have been using heavily. So I know a couple of them are winners, but I'm curious to hear what other people think.

I mean, yeah. What are your computer use experiences here?

I want an app, maybe you can help, that creates a cup of espresso and delivers it to me. Nice.

Yeah.

Does it can be digital?

Physical interaction. Oh, no, no, no.

Some things you just can't do with an app.

Confined to the tray. Yeah. Stay in the tray.

Laws of physics.

Yeah. I mean, there are apps called caffeinated and espresso that are all about keeping your screen awake, but it's more of a digital caffeine, you know?

Well, and there are robotic solutions to that, but you know, it's an awfully expensive cup of espresso. It is. It is.

Only the first one.

Carl, I think you just have to work from a coffee shop.

That's what it is. Well, there you go.

See, this is a.

I mean, if you had this studio, would you be going to a coffee shop to work? I mean.

Probably not.

We should take a break.

Yeah, let's do that. We'll be right back after these very important messages. Developers are asking a new question. Do we still need third-party libraries when AI can generate so much code? For common UI tasks, maybe. but not always. But deep functionality is different. Document editing, PDF generation, rendering, conversion, and enterprise document workflows are the result of years of engineering, extensive testing, and millions of real-world deployments. AI can generate code, yeah, but it can't instantly generate decades of proven document technology.

TXTextControl gives developers that foundation so you can focus on building your applications not rebuilding a document engine. Learn more at textcontrol.com.

And we're back.

It's.

NET Rocks. I'm Richard Campbell. That's Carl Franklin talking to our friend Joe Finney about Traytober.

Traytober.

But you're right. There's so much stuff in the system tray we don't care about. We should put stuff in the system tray we do care about.

Yeah.

Uh, the, the imagination part of developing has grown to be the big dominant part of brain power right now. Yeah.

Right.

And so, and we're not used to that. We're used to, oh, well, if we only had something that did this, okay, we'll get to work on it right away, you know? And now it's, it's you, your imagination is what's going to, uh.

I've been talking to organizations that have gotten through their feature lists on their key products. They're at the point now where it's like, we have to wait for customer feedback now. We outran the customers. And they're not laying off. What they're doing now is going through the back catalog. What have we never gotten to? And also, what never made the list because it was never going to get to it? It was going to end up way back on the backlog. And stuff's emerging. Just so many different bits of code. It's the same thing of, You knew what the barrier was to entry for that idea. So you never went past the idea. And now you got to sort of dust that brain off. It's like, what are all those ideas? Because we could do them. Yeah.

Yeah.

I go into that all the time at work where people, you go into a meeting and talk about hypothetical designs or hypothetical solutions.

That takes more time than just building it.

Let's build it four different ways.

Yeah.

And then go into the meeting and think about it.

You're exactly right. I had exactly that experience. Four ideas for how to solve the problem. Build them all.

Build them all.

And then think about it.

And then showed them off, decided three of them were worth going to ABC testing and put them out.

And I think that's another bigger discussion here. And I'm sure that we, I don't know how much you've been paying attention to the likes of native development news. But with AI coding, you can build for a native now. Previously, a lot of these cross-platform tools and web-based tools was a matter of speed and harmonization and consistency.

Reusability.

Yeah.

Reusability.

Guess what? That's not a problem anymore. That's not a barrier anymore. And that is kind of my point here. Like, I feel like I have too many points with this, uh, this tray experiment here, but a big one is like, look how easy it is to build small native experiences that are nice and delightful, right? Like you, there's all these little paper cuts and like painful points, uh, around your operating system, around the, the programs that you use that it's like, Yeah, you don't want to rebuild Photoshop in Rust. Not that anyone would ever try to do that.

That crazy project that people are doing.

Who would do that? Nobody's going to do that.

Yeah.

Somebody's doing that.

Someone's doing it.

Have you found any models that will write in assembly language or C?

No, I have not. I haven't even tried. I've thought about it because there's... I don't know if you know about the Windows Rust. You can basically build Rust Windows apps.

Yeah.

So.

I've thought about building one of those. One of the apps coming down does use the new reactor UI. That's more of a JSX, uh, react style fluent language. So a lot of it is let's just experiment and see what works. Like let's highlight how easy it is to build native. And when I say native, yeah, I don't mean assembly because the models are really good. Like there was previously some challenges with XAML where they, you know, Wouldn't get demo. They would maybe use more HTML styles and then it would break and fall apart. And then there was no feedback loop. Right. But now there's like this full feedback loop. I can literally give it, I can hold down the space bar, give it a detailed prompt.

I've built a whole skill that explains, you know, what's good, what's bad, how to do it. I give it an icon file. I walk away 30 minutes later.

Yeah.

Right.

We're not even talking hours because this is a small thing. And I have an app that does this little thing.

Mm-hmm. And often these models do a better job with small bytes anyway. Right. And then you can scale it up, like get the initial idea down and then flush it out. More and more I'm feeling like the narrower I can make the spec, at least initially, the better off I am. Absolutely.

And obviously for these, I'm making these little apps. But one, they could grow up, if you will, into a full-on big app.

Sure.

But also, if you are... if you already have a Windows app and you're thinking about, you know, it's kind of bulky and you want a little bit of a utility so that your users can interact with it in a smaller, more like lightweight way, it's very easy to build that experience. It's very easy to build it as an experiment, as a thing you could do in an afternoon, put it out there and see if people use it. Again, you can have all your ideas in your conference room and all your meetings, but you can also just ship software and see if people use it and see what they say and see what they like.

Right. Absolutely.

And that's the whole thing is this, You can get out there and take the feedback and decide, you know, where to even just take it down.

It's not so precious anymore.

Yeah.

Right. Yeah.

The precious part is something for sure where I'm like, I think we need, I have another mobile app that I'm like, I just need to stop being so precious about it and just ship this stuff.

Right. Yeah.

Obviously I need to be careful around API keys, you know, like make sure that I'm not doing anything inefficient.

I mean, ship properly. Yeah.

How do you go quickly?

But also, you know, you're not, you're not, you could do more beta shipping, right? You could just say, Hey, this is pro. you know, try it. Product A, product B, product C, they all do the same thing. Which one do you like best? Rather than, you know, we get everything ready and then we.

Ship, you know, this is, this is apps as cattle versus apps as pets.

I was just thinking that.

Yeah.

That's not pets.

Yes, for sure. No, it really is though. I'm like, I don't, I do give these names, but I don't remember them.

They all start with Trey.

A lot of them start with Trey. A lot of them start with Trey. And another big part of this is when we think about Windows and what Windows will look like in the next 10 years.

Yeah, no kidding.

Microsoft is obviously paying attention to what people are doing.

Yeah.

And the Trey has been this gutter that they don't want to touch and they don't want to really acknowledge, except for the fact that every single popular Windows application uses it. Spotify, Teams, Zoom. Yep. like Outlook, they all use this tray area and it hasn't really been updated since what, Windows XP, Windows 95?

Basically, yeah. And often the innovation for Windows is outside of the tray, right? It's the widgets. It's sort of a proof that they don't trust their own stack either. But I certainly, you know, I spend time on Windows Weekly these days too. So we've been talking about the new leadership at Windows, Pavan Davaluri. And he seems to be listening to the engineers.

Yep.

There's been a lot more work done on stuff that needed fixing and so forth. So you have to wonder. It's been hard to be positive about Windows for a long time. But since Bavan's come on and immediately got attacked by the public for don't ram AI down our throats, just fix problems. And his answer was, okay. Which is astonishing.

And I'm paying close attention. That's really interesting. It's going to take a year. Or so in the pipeline. But I'm also hearing like old engineers are coming back to work. Folks have thought they couldn't fix things anymore.

Yeah.

I mean, the announcement that just happened recently, they announced the new search experience in Windows where you can kind of type queries and quickly toggle settings or quickly do things. And I'm like, great. I love the ability to have these small snackable experiences that are just like pop in, pop out.

You do it.

You don't like the fact that it's like, no, you have to click the start button. Then you have to click the settings. Then you have to click and then you have to click and then you have to click. And now you have the thing you want. And you're like, what are we doing? Like, do I have to click 10 times?

Why are you hiding this? Yeah.

I mean, I get they have to have a catalog, but can you make it, if this is a thing that I do frequently, and I think that's the kind of the thing that windows i hope that they know and i do believe that they a lot of people there know it but windows is kind of our thing like it's owned by the individual people who use it and we customize it we tweak it and we make it work our way where apple is like no you work our way right apple is very much a.

You are holding it wrong yeah.

You're holding it wrong and windows is like hold up however you want and part of that to me is i'm like let's i i'm curious to hear if anybody out there is like really they care about the tray or they've been thinking about the tray or like what could happen there i mean right now we have that little system control that popped up and i'm like what if you could have third-party integrations into there yeah what if you could have but like not just like teeny little widgets but actually like apps like regular apps i i don't know i think there's a lot of opportunity there.

It's interesting yeah i use the bit focus companion a fair bit and it's a little you know it's normally hidden but you can pop it's got a system tray thing you can pop that up and when you and there is a button that says launch gui And that's when they sort of app part appears but the rest of the time it's just in the tray just do a couple of notifications and things that's it yeah that's cool yeah it's great thinking.

And you can do some pretty interesting things i know they don't have like real apis for it but i have done basically animations so i made a i made it out that's not yet released called tray atc air traffic controller nice and it gets air traffic data And as a plane flies over, it will scroll the call letters across the little tray icon.

Nice. So geeky. Yeah.

Blinky lights.

If you hear a plane, you can look down at your tray and you can see what plane it is. What plane was that? And click on it. It'll show up on a map and it'll show you a little radar. And that's the kind of stuff of I'm generating a new icon and replacing the notify icon every single time to create that animation.

And you just create, you're using FlightAware and some. and some geolocation stuff to figure out where you are. I just showed you a view out my window of the ocean. Like I want the same thing for boats going by.

Yeah.

I want a little, a lot of those have AIS. Yep.

I want a little pop-up to do, to manage my smart lights. I want a little pop-up to do my smart home stuff. And maybe it's all from a vendor, but like, why not get it from a vendor?

Like, sure. Dude, do it in a home assistant. come join the cult.

Yeah. Before I asked you if you had, you know, to create one tray thing when, when you pop up, you could have, you know, your applications plugged in there and select them. But are you talking about the, you know, the show hidden icons up arrow, and then you see all the, you know, the services and things like that.

Is that where you're talking about? Are you talking about the tray icons? Like I have for volume and for network and, microphone in use yeah so like where are you talking on the.

Tray yeah so those ones are like the system ones and we're you know developers are not allowed i'll say this we're not supposed to touch but it'd be nice wouldn't it yeah wouldn't it be nice yeah but yeah in that little overflow menu that area is like where all those little icons are that's where the notify icons live yeah all right is in that in that tray.

So the problem i have with this is i don't I haven't memorized what the icons look like for all these 16 different apps that are running in here.

I'm at 20.

And so I know I can recognize Synergy. I can recognize OneDrive and a couple of others, but I have to like hover over all of them to find what I want. So what I'm asking you is if you could just have a tray launcher, that is one icon. And then from there, you branch up and you find all of the apps that are installed and you list those in a menu. Would that not be smart? You indicated somehow that you're already doing that.

Oh, I tried. I tried. But there are no APIs to do that.

Really?

Yeah.

I think part of it is the history. And I think this is kind of where Microsoft is just loaded up with scars of people doing bad things and people who, developers, ISVs, who just, you know, put things in there, move things around. And now it's very locked down. And so there is that like overflow.

So they all have context menus, but you're saying they have to be compiled. Those menus have to be compiled into the, into the icon things, the apps.

Yep.

Yeah.

They're tied in. And obviously I'm not going to say you can't do it. Like there might be APIs that allow you to do it, but, From my point of view, there are not documented APIs that Opus 5.5 knows about as a way to basically manage all of the notify icons. And you cannot move them in and out of that menu.

You can't show and hide them. You can't do the things that kind of a train monitor would want to do.

I'm not sure we're talking about the same thing. So let's say I have OpenVPN, right? I highlight over it says OpenVPN. I right click. Now I get a pop-up menu. connect, view log, edit, config, import file, whatever. I'm saying that would really be great if that could be dynamic. And if we could get that data from somewhere and show it on demand, because then you would have one icon with a context menu that had all the apps in it.

But you're saying there isn't a way to do that.

From what I have researched, there's not a blessed path. There's not a happy path for being able to enumerate over all the try icons and get information about them and manipulate them. It's basically like you have your app tray and that's only yours. And if you want to go look at other applications that maybe have context menus and information and tools capabilities, that's not great.

I don't think that's what I'm asking. I think maybe we're not talking about the same thing.

I mean, are you thinking about an application that manages the tray for your VPN app and for your.

No, that's not what I'm thinking. I'm thinking you have an overall... application that goes in the tray and it's just one and it's, you know, Joe Finney stuff. Right. And when I click on that, right, click on that icon, a context menu would come up and all of the apps that you have written show up as links on that, on that context menu, the battleship gray context menu.

Yeah.

You can, so you don't litter the tray area with lots of icons. If you're building an app a day, that's going to, be very confusing very quickly.

Right.

Yeah, I could build one to manage to basically just launch them, but as far as owning and manipulating their controls, not at all. So it would be a little mini launcher.

But at some point you're getting, you're recreating the start menu too, right?

Right, right.

The real debate here, Joe, is should it be in the tray or should it just be an app and show up on the taskbar?

Right, I mean, if you're going to do a tray app a day, that could overwhelm me, certainly.

Yeah, the experiment isn't that there's 31 try apps and that every user will use 31 try apps simultaneously. The idea is that maybe two or three, but what are those two or three? I think this is kind of the back to the, uh, the discussion around discovery and understanding what is valuable and what is not valuable. I will not be using 30 of these try apps simultaneously.

That's for sure. But I, there's probably five or six that I'll use at work and maybe five or six that I use at home. And it's different. It's a different cross-section of apps. And I don't always know which ones those are. So I think that's kind of the thing of, it's very easy. And it's also a thing of, you know, this is better reach for my phone. So let's see if that experience works. And if it doesn't work, then I don't, you know, whatever, experiment over. Maybe it works for somebody else.

Right.

Yeah.

That's what the nature of experiments.

Yeah.

Well, Joe, this has been very enlightening. And I always love talking about how AI has changed the way we think and the way we develop. And it's, Clearly, you found a little niche that you're using it for, and I think it's great.

Yeah, it's a blast.

All right.

Well, thanks for talking to us, and we will talk to you next time on. NET Rocks!. NET Rocks!

NET Rocks is brought to you by Franklin's Net and produced by Plop Studios, a full-service audio, video, and post-production facility located physically in New London, Connecticut, and, of course, in the cloud, online at pwop.com. Visit our website at dotnetrocks.com for RSS feeds, downloads, mobile apps, comments, and access to the full archives going back to show number one, recorded in September 2002. And make sure you check out our sponsors.

They keep us in business. Now go write some code.

See you next time.

Building System Tray Windows Apps with Joe Finney

Kapitelklicken springt direkt hin

Shownotes

Transkriptvom Podcast bereitgestellt · Zeitmarken springen hin