[Deno] Path to stabilization

[lishaduck]

Current project status

#1 (comment)

Code Status

I plan on eventually PR'ing this to the main effect repo. Here's what I feel is needed first (TL;DR: Full Node compat):

Fill out DenoContext with Deno-native layers (Use of the Node compat layer is acceptable):

• Path (jsr:@std/path)
• CommandExecutor (Deno.Command) [🏃‍♂️ Add subprocess module #3]
• Terminal
• Worker (new Worker())
• WorkerRunner (MessagePort) [Add WorkerRunner and tests for workers #5]
• FileSystem (Deno.* + jsr:@std/fs) [Add a Deno-native file system module #2]

Fill out additional documented platform primitives:

• KeyValueStore (Deno KV) [Deno KV #12]
• KeyValueStore (indexedDB) (There aren't any server implementations of indexedDB anywhere)
• KeyValueStore (localStorage/sessionStorage) [Add localStorage and sessionStorage APIs #11]

Fill out HttpApi support:

• HttpClient (fetch())
• HttpServer (Deno.serve¹)

Ensure support for important² ecosystem packages:

• CLI (Just needs DenoContext?)
• RPC (Needs HttpServer)

Later:

• I haven't looked closely, but Permissions would probably be good to implement?
• Deno is adding otel, we should investigate integrating it

Footnotes

1. Should figure out how to support "declarative fetch"/deno serve via toWebHandler ↩

2. Read "packages I want to use, but would rather wait for Deno support first." ↩

[lishaduck]

Current status: DenoContext is almost done. I'm copying over the tests from the Node platform, finding bugs, fixing them, and repeating. Then, I'll just have DenoTerminal before I've finished milestone 1.

[nounder]

@lishaduck How is work in HTTP area? I'd like to help there.

[lishaduck]

@lishaduck How is work in HTTP area? I'd like to help there.

I've been busy with school and AoC for the last week, so I haven't gotten around to finishing work in KV; even then, I was planning on filling out the Terminal API and finishing up CommandExecutor first: in other words: work in http is lacking.
Feel free to send in a PR! I'd recommend trying to port HttpClient from effect-platform-browser. Speaking of which, I need to update the license to attribute Effectful Technologies...

Also, is there anything I can do to support effect-deno? I didn't look closely, but it looks like compat patches, so if there's anything I can do to help support it from over here, I'm happy to help.

[lishaduck]

Quick update: AoC took up most of my December time, and now I'm on winter break and trying (and failing) not to code the time away. I'm currently trying to resolve the test failures in #15, if anyone wants to give that a shot, I'd much appreciate it.

[nounder]

Pardon for late response. The original notification got lost in EOY madness :)

Also, is there anything I can do to support effect-deno? I didn't look closely, but it looks like compat patches, so if there's anything I can do to help support it from over here, I'm happy to help.

effect-deno is an attempt to use deno's precompiled JSX and SSR without additional build step while running wasm version on vite in the same process during development. It's already wrapped up and ready to be used in first commercial project this month.

Next step is to publish deno-native package (ie. serve minimally modified TS source code) of all Effect packages.

I'm currently trying to resolve the test failures in #15, if anyone wants to give that a shot, I'd much appreciate it.

Looks like @pixeleet nailed it :)

in other words: work in http is lacking

I have finished partial Deno implementation for both HttpServer and HttpClient in one of my project. I will revisit existing impl in this repo and attempt to incorporate it after I come back from holidays.

[lishaduck]

Pardon for late response. The original notification got lost in EOY madness :)

Yeah, I understand the feeling :)

Also, is there anything I can do to support effect-deno? I didn't look closely, but it looks like compat patches, so if there's anything I can do to help support it from over here, I'm happy to help.

effect-deno is an attempt to use deno's precompiled JSX and SSR without additional build step while running wasm version on vite in the same process during development. It's already wrapped up and ready to be used in first commercial project this month.

Cool! It doesn't look pushed, though. All I can see is a patched version of effect.

Next step is to publish deno-native package (ie. serve minimally modified TS source code) of all Effect packages.

Oooohh. That will be nice.
Really makes me want Deno to get overrides support.

I'm currently trying to resolve the test failures in #15, if anyone wants to give that a shot, I'd much appreciate it.

Looks like @pixeleet nailed it :)

Yup. Thanks again @pixeleet!

in other words: work in http is lacking

I have finished partial Deno implementation for both HttpServer and HttpClient in one of my project. I will revisit existing impl in this repo and attempt to incorporate it after I come back from holidays.

To be clear, I've yet to start on an impl here, I was focused on other work.
I'd be delighted to get one, I just haven't started yet. Thanks!

[pixeleet]

stuff like #15 should get trivial with deno outdated --update --latest in the future on the other hand I assume we'd eventually love to see platform-deno merged back into the upstream effect repository therefore I guess pnpm update is just as good?