Cannabat

Hi Will,

I was about to try opencode after using claude code for quite a while.

I think understand the fundamental difference in how they work (acp against existing agentic loops with toad vs a single agentic loop for all models with opencode) but I’m curious why we might want toad over something like opencode, which lets me use any model under the sun.

I suppose toad gets to use the highly specialized agentic loops for each cli. And has a nicer (? opencode is pretty slick from my brief usage…).

Curious to hear about why you chose to built this way and what advantages you see.

My life hack for switching fonts - buy a Berkeley Mono license so I cannot possibly use any other font bc I’d be wasting money. Nice font too.

Very interesting that discord also effectively evolved out of a failed game, albeit a few years after slack.

I'd love to use this but there appears to be no coverage of my country. Any plans to expand coverage to other regions?

I've always been pretty quick to see the hidden image, but I could see these faster than ever - almost instantaneously, even the first one. I wonder if it's something to do with me having stared at a screen for a decade since the last time I saw a stereogram...

The article mentions it, but you can sum up print debugging as selectively enabling verbose/trace logging. “We’re about to do X” or “We just did Y, here is Z”.

A debugger gives you insight into the context of a particular code entity - expression, function, whatever.

Seems silly to be dogmatic about this. Both techniques are useful!

Unfortunately, it’s not possible to express this sentiment via election participation. Abstention ends up supporting one candidate more than the other. What seems to be an affirmation of neutrality is not that in practice.

I remember using cash about 10 years ago. Was it under a different user back then? Ken wheeler maybe?

Thanks for your continued work on it!

The team who create Atom, Electron and tree sitter are working on zed: https://github.com/zed-industries/zed

3 to 5 times more valuable is a huge leap. How do you measure this? Are you a software engineer?

I used RockAuto for the first time recently. The site is super dense with info and functionality, but never once did I feel lost or overwhelm. Neither did I feel like anything was missing. Everything I needed to see and do was exactly where it should be. It's so rare for the screen to disappear like that. It's so transparent that I _almost_ can't say it's a joy to use (it's a joy to use!).

Thanks for for the detailed answer. I'll try it out some time :)

Knip does have some auto fix capabilities but it’s not perfect yet. The detection is great, though. I use it with `dpdm` to keep things tidy. With these (plus eslint, prettier and typescript) in CI, I feel all warm and fuzzy.

OP, can you describe differences from knip?

Urine is not sterile [0], but it probably won't hurt you either.

[0] https://pubmed.ncbi.nlm.nih.gov/25766599/

Very cool! Did you explore using the gap properties to specify the thickness of your boards?

I wanted full control of the design process. Doing it in vector illustration software was too painful and it was taking too long.

Aaaand then 5 evenings later… classic :)

I find myself wanting a result type so often in python. This looks great. Thanks!

Thanks - I was looking in the wrong place in the docs.

I wonder if renovatebot [0] supports this?

[0] https://github.com/renovatebot/renovate

I don't know if I'd use the word "enjoy", but it certainly makes python tolerable. With all due respect, it blows mypy out of the water.

Same. I'd hope that using LSP and Copilot would not meaningfully impact insertion latency, though - I'd expect those to be fully async, in a separate thread.

These are tests of the latency between a keypress and the display of a character in a terminal

Wow, the numbers are entirely unexpected. I want to see the latency when you have no display server, having booted directly to a shell.

I imagine modern competitive twitch-reflex videogames (fps, fighting) must have far less input latency - 100ms must be totally unacceptable. Suppose that's the power of a GPU.

respond to your keystrokes on the next display refresh

Claimed insertion latency for a keystroke: - zed 58ms - subl 75ms - vscode 97ms - clion 83ms

60 Hz displays refresh once every ~16.6ms. 58ms to insert a character is ~3.5 frames. My laptop can do 120 Hz, and other monitors can even do 300 Hz.

Yeah, it's a lot faster than the other editors, but 58ms seems really slow - definitely not "next display refresh". All of them seem really slow.

If you want to implement the suggested changes across your codebase, you can use the `flake8` plugin `flake8-comprehensions`, or `ruff`'s `C4` ruleset, which is an implementation of `flake8-comprehensions`.

Always throwing away the first draft is such a useful strategy. Besides almost always resulting in a better finished product, it keeps you humble and light - life is an experiment!

Check out the album cover of Sheath by LFO (as mentioned on the page). Great IDM album.

Even if the benchmarks are accurate - or even overly conservative - it's possible that these are the "uncensored" versions of the models, and what the public gets access to in the end is far less capable.

That's a simple and practical solution to a common problem. Thanks for sharing.

Yeah but it’s hard to find the hardware this OS is designed for. There’s little need for it. Compared to the oodles or low powered arm boards out there that would love this. So the missed opportunity is just that a lot of effort has gone into a very narrow platform that would probably be more appreciated on a diff platform

Super cool project though of course

Those devs _can_, but why should they? Code formatting takes a lot of time - it doesn't matter how skilled of a dev you are - you can only accurately <tab> and <space> so fast.

But the big win for code formatters is fewer decisions. Decision fatigue is real.

What's wrong with the application of e.g. Prettier? Is it wrong enough that it's worth your time to manually format however many thousands of LOC, making countless micro decisions along the way?

Do you plan to maintain the FF Replay fork?

Great work Replay team! Awesome tech.