MintPaw

I don't think that wasm is inherently uncompressible. I think it's more that tools like emscripten include the entire C++ standard library because most existing libraries expect them to just be available for free. But hopefully that would change as more web focused libraries are created.

I'm surprised no one really has answered the implied question. Wireless protocols are hard for VR not because of bandwidth, but because of latency. Head tilt motions have to be processed extremely quickly to not be nauseating. Even with wired systems, the headset often ends up doing all kinds of motion prediction to process motion and shift the image before the game has time to process it, because engines aren't going to round trip input to video output at 120fps/8ms (double buffering doubles your latency!).

Of course, that multiplies with bandwidth requirements, which are also quite extreme at 4k 120fps.

It doesn't seem to me like you've described concepts that are all that different, loading from global memory is slow on the CPU too because often it requires some kind of barrier or lock. All techniques require you to pre-arrange memory, just in different sized blocks. And sharing data between "lanes" is always to be minimized. Different amount on different platforms ofc, but the ideas seem similar.

Doesn't this break if the child uses a VPN, or some other software to spoof? For this to really work Windows would have to lock down all HTTP, and really all networking, only allow approved software/browsers, etc. Is there a reason this wouldn't be the case?

Why would this be true? The way "single instance mode" usually works is by taking a global mutex or making a lock file. What makes this not possible for TUI apps? Isn't it just culture?

Probably because it makes distilling and external analysis/tracking harder.

It's an interesting thought, but hard to imagine how this would be implemented system wide. The clipboard would need to remember the source application, and have some kind of unique identifier to issue a delete to the source application.

I see your point, but this isn't exactly true, it only takes a single person to bias a model by deleting specific training data or over training on certain facts.

I'm not sure what the situation is here, but on many sites making alt account to avoid bans can result in a lifetime ban. A high profile figure who's been banned from a service advertising that they still have access to it might be a risk.

Also, these companies (or data brokers) likely share ban and alt account lists, ban evading could hurt the reputation of all your accounts.

These are some pragmatic reasons not to evade bans.

This is a general problem with destructors, you can't "batch delete" objects. To free a lot of stuff you're required to go pointer by pointer through the tree to clean up each object. To get real performance gains from pools you can't have per-object/subobject custom cleanup code.

This is way too absolute a point. There's services like water and internet service you're effectively subscribed too that don't get worse. There's obvious efficiencies that can be gained from regularizing demand with subscriptions. Amazon Prime is another example of a something I've been subscribed to for over a decade and I don't feel has gotten worse.

Not you particularly, I meant the original: "it feels like every share of income is at its lowest except for the ultra wealthy."

It's ambiguous in several way, no time scale, "ultra wealthy" isn't defined, and "income" somewhat ambiguous.

Data is the answer, it's just that so few people are willing to look at unbiased data. Although the start is asking a more measurable question.

There's a whole category of video content called "First Amendment Audits" where people film with big cameras in public and farm people random, and often quite hostile, reactions.

This is at odds with the idea that devs should take full accountability for LLM generated code.

He also didn't even say the code was generated by Claude, just the description, although a decent amount of the code probably was.

A side note, isn't package maintenance something that can actually be solved to some extent by LLMs? The prompt would be something like "Clone this repo and build this package while building/bundling as few other packages as possible with minimal code changes."

Then set it in a loop on all the packages for a particular system, I don't have experience in package maintenance and would be curious what kind of issues would come up.

I'm not sure about that, they could do far worse than delist you if they really hated you.

Interesting point, but I'd always thought the opposite, you're much better protected by the law if you use services from your own country.

If you use a service outside your country, I believe you could have all your code stolen and get hacked/exploited in a way that would be totally legal.

What is a voicemail in this context? What app is reading it?

Detect the mouse moving in a "non-human way"? If it were that easy there'd be no hackers. And even if it where, what about wall hacks?

There's just different types of programming, I also find that if I give a decent description of a bug an LLM will often find the problem, and that great in a system that's mostly legacy and hard to repro. LLMs are also good for quick small scale from-scratch projects.

But there's a middle ground where you're really have to build something out that's super complicated and performant. Or do refactors that have a high quality bar. Situations where code bloat really matters. LLMs tend to create crazy amounts of code, not really thinking through the broader system and taking system level in-variance constraints seriously.

I recently had my faith shattered with I saw someone lock onto an ally through a wall in a kill cam, and I haven't played sense.

Blatant cheaters are bad in some ways, but subtle cheat are far worse imo.

I think there's a large subset of programmers now who consider null checking (or even the existence of null) to be bad, and prefer something else like exceptions or option types. I don't get it personally.

$600/mo? Do you mean $600 as a one time purchase for life? I've never heard of any Adobe plan that expensive.

This looks to me like the issue is that Zed is using too many buffer objects, it should be querying the Vulkan context to see what's the max and sticking to it. So it seems like Zed's not doing it right, or maybe the AMD driver is failing to report it correctly?

It is a problem though, the GPU apis are pretty terrible. But with such large modern displays it feels important to have a GPU accelerated path. Maybe sticking with OpenGL would be better.

Vulkan, DX11, or Metal, I'm curious which environment has the ability render a GUI desktop but doesn't have access to a modern renderer pipeline. VirtualBox I guess?

Everyone talks badly about Cursor and it is kinda a piece of junk, but no, there's nothing that has the features of: being able to see agent diffs in an editor, seeing diffs inline in chat, be able to click them to jump to the code, and being able to click old chat messages to edit/fork them.

Those are basically my only requirements, and it feels like I've tried everything and they're all everything only has 1 of those features. Zed is the closest, it technically has those features, they're just buggy and have provider specific quirks.

So I'm stuck on Cursor until Anthropic invents IDE technology, or at least VS Code wrapper technology.

Yes, but I got a subscription because I was tired of alt+tabbing to the Cursor spending dashboard between prompts to make sure I wasn't over spending. I'm ok if they slow me down for a few hours during peak usage. But getting cut off for 20+ days because I'm not thinking about the prompt cache for a bit makes a subscription feel pretty useless.

I was using it with Zed before, because I guess I'm one of the only programmers who doesn't just full vibe, which seem to mean I'm not the target customer for a lot of these companies who seem to be going all in on the terminal interfaces.

I've gone back to Cursor auto the last few weeks, it hasn't been too bad actually, I haven't managed to run out of the $20/mo plan yet.