4,679 karma · joined September 24, 2013
lab notes: untested.sonnet.io about: sonnet.io
mastodon: @raf@mastodon.cloud twitter: http://twitter.com/rafalpast email: hello@<my website domain>
important meeting notes: https://potato.horse
Why?
Also:
> Signal has been a primary method of communication for federal workers looking to blow the whistle on DOGE. > from https://www.disruptionist.com/p/elon-musks-x-blocks-links-to...
Btw, I don't live in the US but I sketched a simple tool to prevent X from censoring Signal.me links: https://link-in-a-box.vercel.app
Oh yeah, 100%. One of the reasons I like Svelte are colocated CSS modules.
> seems like frameworks are somewhat converging now on this app router concept.
That for sure, and signals in one way or another. It took us 10+ years, Solid.js then preact to bring back a pattern from 2010s (IIRC) implemented in Knockout.js.
People are talking about declarative signals in HTML. I don't have a formed opinion about them yet.
const state = new Proxy(
{},
{
set(target, property, value) {
target[property] = value;
renderUpdate();
return true;
},
}
);
I'm only 1/4(.0000000004) serious here. That said, I've been doing this for almost 30 years, I was a kid when I picked up Netscape, I have gone through every mainstream and many niche libraries/languages replacing libraries. Now, in my late 30s I finally can grow a moustache (a big achievement in my family gene pool) but the analysis paralysis is still making me anxious every time I pick up a new project.- preact w. signals or Svelte are still my libraries of choice if I need to spike something, but that's partially because of signals and vite. The development experience is good, there isn't much magic you need to deal with on a daily basis.
- Web Components are great in many ways, but they don't solve the problems I usually have with bigger projects (scaling state, orchestration for the lack of a better phrase). Lit makes it a bit better though.
- JSX is actually not that bad, as long as you treat it as a DSL around functional composition. Native JSX + signals would probably make most of my node_modules redundant.
(please don't overthink this, I didn't)
- you could render the page using puppeteer server-side, getClientRect/calc and apply the dimensions to the path, then spit back the markup, OR
- you could use HTML + CSS to render the bubbles
People already started using this thing, so if there's an issue with it, I'd like to respond/apply fixes if needed.
https://link-in-a-box.vercel.app
If you think someone might benefit from it, please share. Also, spam if you have feedback!
A little bee told me that VLC + most torrent trackers are still a better UX than most Netflix/Apple TV/Prime video.
None of them will crash after 30s of playback because your postcode changed for a week, because you have an ad blocker installed, or because you just plugged in an external display and some DRM magic kicked in. Just find the file, download it and double click.
Aaand, here's another one I like: https://days.rory.codes
(not from the US obv.)
Yeah, it should be! I'd say just give it a go and see. CO has also some interesting workflows for managing photoshoots, a more classic one similar to LR and a session-oriented one. The bulk update UI is not as nice as LRC imho.
I remember hearing something about Darktable splitting into two separate projects, one with a better UX? (I'm probably wrong re the details, on the phone atm)
Ironically importing 40k or so photos from my old catalog into CO took just a fraction of time needed by LR to do the same. That was one of the triggers to stick with it for a few more days.
I had been using Adobe since Photoshop 5.5, but ditched it a couple of years ago and I’ve never looked back.
https://untested.sonnet.io/notes/xitterpng-privacy-friendly-...
(this is a response to the comment, not the article)
I think it’s a good balance between locking the user into your product and dealing with the cost of a constantly evolving service.
1. I need a smart autocomplete that can work backwards and mimic my coding patterns
2. I need a pair programming buddy (of sorts, this metaphor doesn't completely work, but I don't have a better one)
Pair development, even a butchered version of the so called "strong style" (give the driver the highest level of abstraction they can use/understand) works quite well for me. But, the main reason this works is that it forces me to structure my thinking a little bit, allows me to iterate on the definition of the problem. Toss away the sketch with bigger parts of the problem, start again.
It also helps me to avoid yak shaving, getting lost in the detail or distracted because the feedback loop between me seeing something working on the screen vs. the idea is so short (even if the code is crap).
I'd also add 5.: use prompts to generate (boring) prompts. For instance, I needed a simple #tag formatter for one of my markdown sites. I am aware that there's a not-so-small list of edge cases I'd need to cover. In this case I'd write a prompt with a list of basic requirements and ask the LLM to: a) extend it with good practice, common edge cases b) format it as a spec with concrete input / output examples. This works a bit similar to the point you made about generating unit tests (I do that too, in tandem with this approach).
In a sense 1) is autocomplete 2) is a scaffolding tool.
So, I started untested (https://untested.sonnet.io) where for 111 days no matter what I'd share a post per day. Here's the first post: https://untested.sonnet.io/notes/111
Consistency really helps - there are posts that I had spent weeks on and barely anyone read them in the first place. Then, once I started posting smaller, messier notes almost daily, both the traffic (not that important to me) and the amount of interesting interactions with people (very important to me) went up by 100x.
Another way in which I avoided yak shaving/scope creep was to ditch my own site and rely on Obsidian Publish (I went back to my own solution after a year, as a reward for mostly sticking to the plan)