warpfork

if someone says they want to protect the children and their policy proposal for doing so doesn't involve giving children more rights, more agency, and more control over their own lives and property, you can usually assume they're either lying about their motive or being lied to about their motive

I get yelled at a lot on BS for saying so, but this is 100% true. There are people overhyping AI, but the alternative is not that AI is useless, or even the average of the two positions.

A lot is going to change dramatically even with today's AI. Ignoring that means no chance to shape what's next

I am MONSTROUSLY HYPE about this.

- the web is the best ui platform we've got
- the web desperately need some way to distribute code and apps that doesn't involve "backend" "servers" and the whole split power dynamic that implies
- ... And maybe this is it???

Web Tiles Tiles are composable web docs and apps that can safely be used in arbitrary contexts and assembled from multiple sources to carry out complex user interface tasks.

Hot off the internets — there's a brand new kid on the DASL block: Web Tiles!

It's still early days, some key features are missing, and there are some rough edges, but you can publish tiles to AT and play with the idea!
webtil.es

Dude you're one of the most inspiring people on the internet.

Go be as loud and proud as you want to be, wherever you want to.

I'm pretty sure lots of people at that event VERY want to see whatever you want to present.

The For You feed will be down for a few minutes.

I need to restart the computer to try resolve Steam segfaults (probably another nvidia driver update?)

Will use this opportunity to enable RAM EXPO profile (4800 -> 6000) which got reset during the CPU upgrade.

Fun fact, playing Overwatch increases For You response time from 90ms to 115ms.

I think the American ethos, correctly represented at the national scale by MMP voting, would be a wildly powerful and wonderful thing to behold.

What we could get with true representation, instead of this gerontocratic circus of gerrymandered corruption, is almost unimaginable.

oh no

so many unexpectedly routeable ports

My mind keeps looping back to this:

Our expertise managing scarcity isn't (particularly) useful managing abundance.

Abundance can be a good problem to have, but it is different than scarcity, and it is a problem.

I think this is true of stuff, ideas, art, news, expertise, code, etc.

I believe I have counterevidence

Congratulations, Doctor 🫡

How many doormen do we have to fire before we stop heedlessly removing humans from systems deemed "lesser"?

And automation only displaces burdens (often in ways not understood until too late), it doesn't solve. Logistics can never be compressed beyond the limits of objective time and space.

Wow, so sometimes targetted agentic censorship is subtle.

And sometimes it's... This.

(If it worked for you, great. I'm just frustrated every time I see it because it seems pretty stalled, at a place that's hard to either secure nor make performant, and... I dunno, it seems like an mvp of the sort that's defacto sapping the energy that I wish was flowing into something better.)

Oh, i didn't realize fyne had a webview binding. That's neat.

Still, webview/webview overall...ooof.

- no way to sandbox network
- no api for performant passing byte buffers
- it's pretty much yeet strings to eval? That's...It?

I don't wanna diss the effort made, but it plateau'd in a bad place.

Whatever the Zed folks are cooking on is probably going to be above-average interest... If it ever gets fully disentangled from Zed itself.

Multi-line text areas are always where widget libraries get catastrophically hung up, istm when I survey the field, so anyone Past That has a good shot.

If you revisit that project, you may wanna check out wails.io . It's got *much* better, richer, and safer bindings than what the webview/webview project offers.

It's not a very "elegant" theosophically rewarding thing to say, but...

I think maybe more research into how to let programmers use arena allocation safely might be worthwhile, and might find some sweet spots for productivity+safety+performance that are currently wildly underexplored.

Memory strategy in programming language design theory is an interesting topic,

because on the one hand, I don't expect any non-GC solution to be reasonably Correct without the level of rigor that Rust shoots for (eg affine types)

and on the other hand, serious gofast is "haha arena allocs go brrr"

Ok, ok, I lied.

It's really an overlayfs with a unique tempdir as the upperdir per boot / kick.

Unexpected state delenda est.

tmpfs

You *can* get angy and descend all the way to code in it (and sometimes i have, to occasionally shove some weirdly specific html into just one little bit).

But you don't have to! drag and drop, or nudge positions with arrow keys, is 100% there, and i use it almost exclusively. 👍

Create and Share Presentations for Free | Slides Slides is a place for creating, presenting and sharing modern presentations. Sign up for free.

slides.com is basically a wysiwyg on top of reveal.js (by the authors of it, as well, so paying them for it is effectively supporting foss). I've had a pretty good experience of it.

Templating could be better last time I jousted with it, but maybe I also just didn't try hard enough (likely!).

bsky.app/profile/did:...
is super good and widely used

bsky.app/profile/did:...
is maybe a bit experimental and I'm not sure if still attended to, but also in the realm

...which is maybe not an intuitive result, but.

I think the whole software engineering field has a serious piece of under-estimated meta-debt in that "code review is harder and more error-prone than code writing". And LLMs didn't create that challenge, but they sure aren't benefiting from it.

My feeling is that code review importance hasn't gone anywhere, and is in fact rising;

And there's actually a substantial number of cases where it's actually easier to just Write The Damn Thing Myself than it is to code-review the LLM output sufficiently thoroughly.

(again "purely" being a super load-bearing word there)

All true but I think

- there was usually a substantial kernel of good-enough software at the historical core of most now-large incumbents, even if their codebases have now "lost the plot"

- i see purely vibecode projects "losing the plot" way too fast, at way too small sizes.

Yeah, we're probably haggling on different meanings of success. I agree that many business units survive with bad code. I consider that more of a "despite" thing.

Newer companies with better code aren't *guaranteed* to displace larger, worse incumbents. But they've usually got a shot.