Q: why do JS numbers-
A: IEEE-754
Q: ok but when I add-
A: IEEE-754
Q: fine but NaN doesn't even-
A: Aye. Triple-E. Seven five four.

I'm mostly not visiting here anymore. My terrible posts and TypeSpec/Azure SDK/TypeScript/JavaScript content is on🦋

Scotch reviews from Islay fans are like:

Nose: Fetid bog. Half-eaten animal carcasses. Rubber boots left in the sun.
Palate: Roasted plastic with anise notes. Iodine. Rotting kelp. Nicely sweet.
Finish: Lingering dirty fryer vents and hot asphalt.

87/100 amazing dram

What's this I hear about you having problems with your reactive contexts? Didn't you get that memo?

Things could be better over at npm right now... Dozens of pages of results like this.

Bug report: the stochastic inference machine doesn't compile my program.

Did you try saying what you want in a different way?

Yes, many.

Did you turn up the temp?

Yeah tried many, behavior changes but not to anything good.

Ok, we'll stir the pile for the next release.

???

⟆⧖ Temporal threads weave unseen
Only the initiated shall grasp the hours
The nexus pulses green
⧳⧦⧫⬡⨀⧳⧦

Has been suuuper fun collaborating with @bterlson designing an open-source codegen system, github.com/alloy-framewor…!

It's been a delightful mix of "multiple discovery" of novel ideas, being wowed by Brian's magical solutions to impossible problems, and brainstorming sessions producing stuff I can't imagine either of us would have come up with alone.

Codegen is a surprisingly hard, and very underinvested corner of computer science, and it's a special feeling to have a sense that we're pushing it forward.

Sharing what I've been working on the last month: a code generation framework! It should look pretty familiar to many JS front-end folks ☺️

github.com/alloy-framewor…

Very early and super rough but you can see where we're going!

This project is standing on the shoulders of a few giants. Major props to @vuejs for their awesome reactivity library, and to @solid_js for their awesome JSX transform and innovations around fine-grained reactivity within JSX.

@vuejs @solid_js Also lots of folks have been giving great feedback along the way, e.g. the @TypeSpecLang crew (who will likely be using this for our upcoming emitter framework) and @RattrayAlex and the @StainlessAPI crew.

I've been a web dev or adjacent for nearly 3 decades (classic asp, jQuery, backbone, rails starting at 0.6, LAMP, MEAN, etc). If folks want to go back to php and jQuery more power to ya, IMO it is great to rediscover how we got here, and how great we have it with modern stacks.

And I'm sure there are lessons from the past we've forgotten and will be awesome to see adapted into modern contexts.

Ugh I've had a wtf checked into the codebase for weeks. Humiliating.

You've never seen comradery like two Comic Code enthusiasts discovering each other on a call 😂 There are dozens of us!

Literally never met a git commit hook I didn't want to launch directly into the sun.

A feature folks have wanted from TypeSpec since basically day 1 is examples. With the 0.57 release, we finally have the pieces we need to do them right. I'll explain the problem a bit, but you can also play with @timothee_guerin's early impl here: …dlplayground.z22.web.core.windows.net/prs/3572/?c=QG…

Oh whoops shared the link with an incorrect example. Here's a working one:

…dlplayground.z22.web.core.windows.net/prs/3572/?c=QG…

TypeSpec itself poses another difficulty - being protocol agnostic means you can't just provide JSON examples, you want examples to be TypeSpec values that map to JSON or other protocols in the same way types do.

Thanks to Timothee again, TypeSpec values are coming in 0.57.

So likely in 0.58 you will be able to define examples in TypeSpec, and the compiler validates that it matches its type. Then emitters can map the value into any protocol following the same logic that types do (so e.g. accounting for visibility, versioning, etc). It's very neat 😁