Ashley Claymore

I don't think there is a mapping page. Under the hood it uses amaro, which is a wrapper around swc.
So to know if certain syntax is supported you'll need to work backwards from when it was added to swc.

Yeah import extensions are the main complexity there. As still nice to be able to run the `.ts` directly during `--test`.

As long as you're using the right versions of Node.js yeah! It follows the same approach as ts-blank-space, replacing type annotations with whitespace.

nodejs.org/api/module.h...

A `package.json` with zero dependencies and the following build script: node -p "require('node:module').stripTypeScriptTypes(''+fs.readFileSync(0))"<index.ts>index.js

If anyone needs a silly, not portable, zero dependency, one liner TypeScript build script that can only handle a single file. Look no further!

The other day I had an AI saying that there was no such thing as Date.prototype.toTemporalInstant.

Even after I gave it the link to the mdn page it was still very suspicious and didnt give me a "you're absolutely right"

Add abortSignal.addAbortCallback by jakearchibald · Pull Request #1425 · whatwg/dom This PR is a discussion starter. The goals: Make progress on allowing TC39 able to adopt/use AbortController and AbortSignal without requiring Event and EventTarget. Provide a system where abort &...

github.com/whatwg/dom/p...

Add Explicit Resource Management update to agenda by rbuckton · Pull Request #2048 · tc39/agendas

'using` isn't stage 4 yet.
If it gets conditional advancement at the plenary in 1 weeks time then it might make it in before ES2026 🤞🏻
github.com/tc39/agendas...

I don't immediately hate that code either.
I would love to start seeing more code unattach their abort handler when they complete - it's way too uncommon today.

Slide 9 isn't using 'using` on the DisposableStack btw

MockMethod memory leak · Issue #41 · 43081j/hanbi Each call to mockMethod creates a long lived restore function that leaks if never used. I'm wondering: might be worth documenting the risk of not calling `restore' after tests mockMethod could eith...

or maybe easier to discuss on gh
github.com/43081j/hanbi...

Looks like creating method stubs is a leak until restore is called.
Maybe worth calling out in the README @43081j.com?

"See this stash, don't try this again - you will lose marks"

Is this the one you're implementing in prolog?

Mutable Signals - Reactivity's Missing Link YouTube video by Ryan Carniato

Hey! I've got this stream queued up to watch next. Looks like it covers stores, nested getters, and diffing.

www.youtube.com/live/jfZdl3Q...

Thanks for the extra info Jovi!

I was reading the signals docs and was a little confused by preactjs.com/guide/v10/si...
The example shows passing in global signal, but looking at the implementation it looks like that would return Signal<Signal>.
I'm wondering if I'm missing something or if the docs are not the best example

Thanks Ryan! Much appreciated. I'll have a dig back though the old streams around that time.

@ryansolid.bsky.social was there a stream where you covered the solid 2.0 createStore ideas?
You talked about it briefly in the more recent stream and it sounded very interesting - only diffing store changes when there is a known listener (I think thats what I heard).
Thanks!

In terms of potential 2.0 'breaking changes' - it looks like there is a strong chance that `setSignal(newValue)` and then immediately reading the signal will now see the old value.

Though I'm not sure how much code is relying on the current semantics. I can't think of a case off the top of my head.

Loving the streams btw! Being able to switch a memo to be async and it can still render without more changes is super impressive.

arr["-0"]; // oh no!

Thanks for looking further into it. I'll have another play to see how I get on 😃

Nice! Will make it even easier to compare 6.0 and 7.0

Let's resolve modules once and for all

Feels a shame to need to make the resource a public field.
I think I'd prefer to explicitly register it locally:

```
# privateField = this.#stack.use(new Resource());
```

Admittedly more verbose.

Pastebin? 🙃

bsky.app/profile/ashl...

I don't think this works in practice. If the components are rendered in a list the indexes that form the keys will go out of sync over time.

Compared to react where the data is "plain immutable objects", when the data changes the whole object changes.
Downside: "Key" is needed to track the underlying "identity" of the data.
Upside: slightly easier to inspect the data when doing debugging/console.log.

'.map' wouldn't work. Solid tracks the references. The expectation is that the objects in the array don't change. If it's a list of Todos then each item is consistently represented by the same object that has inner signals that represents the changes to the item.