> For something like Audacity (the audio program), how the heck does it make sense to put that on a website (I'm just giving a random example, I don't think they've actually done this), where you first have to upload your source file (privacy issues), manipulate it in a graphically/widget-limited browser

I understand it was just an example, but you'd be surprised how far browsers have come along with technologies like Web Assembly and WebGL. Forget audio editing, you can even do video editing - without uploading any files to the remote server[1]. All the processing is done locally, within your browser.

And if you thought that was impressive, wait till you find out that you can can even boot the whole Linux kernel in your browser using a VM written in WASM[2]!

But I do agree with your points about lack of feature stability. I too prefer native apps just for the record (but for me, the main selling points are low RAM/CPU/disk requirements and keyboard friendliness).

[1] https://news.ycombinator.com/item?id=47847558

[2] https://joelseverin.github.io/linux-wasm/

Sure, but taking your video editor example, what advantages does an in-browser app provide over a native application like DaVinci Resolve, other than portability and not needing to install the application, in exchange for reduced performance, a clunkier interface, and reduced integration with the rest of the desktop platform?

And if this is such a compelling value proposition for full-featured desktop productivity applications, why didn't Java Web Start set the world on fire?

Portability and not having to install the app is a huge advantage. Especially on operating systems where there aren't any decent choices. Take Android for example, the Play Store is full of rubbish and adware-riddled apps, finding a decent app in there is like looking for a needle in a haystack. And whilst FDroid exists, most of the apps there are pretty basic in general, especially wrt this example (video editing).

Putting aside the video editing example for a bit, consider the photo editing web app Photopea, which is an excellent alternative to Adobe Photoshop. Linux is in urgent need of a Photoshop-like editor (and no, GIMP doesn't cut it), but Photopea does a decent enough job for many amateurs and even some pros. For a lot of these folks, Photoshop is one of the last things stopping them from switching to Linux, so apps like Photopea fill that gap. And guess what, Photopea works great on Android too.

Another use case is restricted environments where you can't easily find and install apps, eg immutable distros, or work computers. I use Photopea on my work PC quite regularly for light editing, because MS Paint sucks, and my role doesn't really justify going thru the hassle of getting the approvals to get an editor installed. So like it or not, web apps have their place.

> Sure, but taking your video editor example, what advantages does an in-browser app provide over a native application like DaVinci Resolve

It's the issue of friction. Also, good webapps are often _better_ than native apps, as they can support tabs.

> And if this is such a compelling value proposition for full-featured desktop productivity applications, why didn't Java Web Start set the world on fire?

Because it relied on Java and SWING, which were a disaster for desktop apps.

> Also, good webapps are often _better_ than native apps, as they can support tabs

All the native apps I use support tabs, its a basic feature of the macOS windowing APIs https://developer.apple.com/documentation/appkit/nswindowtab...

This is a relatively new API, and the native apps that I use still don't support it properly. There are also things like middle-clicking to open things in new tabs, and being able to bookmark locations.

> Also, note that there's zero CUDA dependency.

So does this mean I can run this on AMD? And on a consumer 9000 series card?

If you don't have the source code then it makes no difference. If you have the weights and are running some model via llama.cpp, then you are using whatever API llama.cpp is using, not the API that was used to train the model or that anyone else may be using to serve it.

If you found a rare 9000 card with 200+ GB of VRAM, sure

If the card supports vulkan and the model has gguf weights. llamacpp has excellent vulkan support that is being actively developed and is not that far behind CUDA where speed is concerned.

* https://github.com/ggml-org/llama.cpp/releases

Luckily, we had web accelerating proxies like OnSpeed[1] back in the day that would compress web pages (including lossy image compression) so if you were one of the poor sods still on dialup (like I was), it was a lot more bearable.

[1] https://en.wikipedia.org/wiki/OnSpeed

Oh neat, I'd never heard of them. Almost feels like a spiritual predecessor of CDNs, serving optimized assets from existing websites via their servers.

It's not impossible, but most KeePass tools are written in sane languages and built with sane tooling, and don't use trash like Javascript and npm. Of course I'm not considering browser extensions or exclusive web-clients, but the main KeePass client has a good autotype system, so you don't really need to use the browser extension.

In any case, the fact that the official BitWarden client (which uses Electron btw) and even the CLI is written in Javascript/Typescript - should tell you everything you need to know about their coding expertise and security posture.

Fully agree, I can't wait for the day when developers finally stop using javascript for shit it was never designed for. .NET is decades ahead at this point.

You need to consider the upgradability aspect too - the next time you want to upgrade, you just need to buy a new mainboard, which would be considerably cheaper than buying a whole new laptop.

A new mainboard that may need new memory. From another comment in thread:

  I have 64GB of DDR4 in my current laptop, and replacing that with the same amount of LPCAMM2 LPDDR5X is probably more expensive than the rest of the laptop itself.

And another:

  I just had my mainboard die, and I was advised there currently isn't another mainboard in stock that works with my old DDR4 RAM

I've never seen it stated anywhere that the App Store is the primary method for macOS. Well, I could be wrong, maybe Apple does mention it somewhere, but pretty much every popular app publisher still publishes their .dmg file directly on their own website, much like most Windows developers.

At least I've never had to use the store in my 15+ years of using MacBooks, and I can't see myself using one anytime soon, unless Apple starts forcing you to (in which case I'll just stick to using homebrew).

They have a 16" version btw.

But these people would already have a company-provisioned laptop.

I thought the idea was the Framework is the company-provisioned one. Or are they letting people do corp work on personal laptops?

I don't want to maintain state on two laptops

With the right tooling, the state is minimal and gets synchronized between laptops.

Why is the AMD version not available in LPCAMM2, and can we expect to see an LPCAMM2 version in the near future?

Probably because it's not really a brand new AMD motherboard?

I'd expect/hope to see whatever comes after Strix Halo in a Framework motherboard. That's when you should be looking for LPCAMM2 for AMD.

Not sure fhe Halo series will be LPCAMM compatible. The framework with the Halo chip is soldered.

I looked into it a little and it does seem like LPCAMM2 just can't provide that memory bandwidth through a single socket and dual socket would be overly complex to wire up. Bummer :-(

Hopeful that's no longer an issue with the Pro version, being made of a machined single-piece aluminum block.