Hi - when deploying a worker using a service binding set up in wrangler, I'm getting this warning:

▲ [WARNING] Processing wrangler.toml configuration:

    - "services" fields are experimental and may change or break at any time.

I can't see anything in the docs about this being experimental - is this something I should worry about?

maybe a better question for the #799437470004412476 channel 🤔

I don't really know why that warning is there

They're GA - https://blog.cloudflare.com/service-bindings-ga/

yeah I guess it's either an old warning that hasnt been removed from wrangler, or the wrangler set up isn't set in stone yet

It's specifically because they're not done in wrangler yet

they only work in local mode, and it's still a little flaky sometimes

it worked fine in prod for me

I'm talking about dev

Hey guys, I'm currently getting this error while trying to setup custom scss builds:

No loader is configured for ".scss" files: src/styles/index.scss

I have a esbuild config that looks like this. When I run the file with node directly it builds, but when I put that same node call in

build.command

I get that error. Is it possible that wrangler is still trying to build it with its own esbuild?

What's your

main

value in your wrangler.toml?

My guess is that it's

src/index.tsx

, not

dist/out.js

I just changed it, still the same error :/

Can you look at the out.js file and be sure that it's not trying to import an scss file?

Nope, its all inlined

Is there some way to maybe clear wrangler cache?

Theres no cache for this

Is there a repo I can look at?

Oh, I solved it by removing the index.tsx path in the npm script, forgot about that, sorry

Well it works now, thanks!

Great 🙂

Does a Worker connecting to remote Web Socket server ever count as "sleeping" while we wait for data? https://developers.cloudflare.com/workers/learning/using-websockets/#writing-a-websocket-client

I setup cloudflare cache api in my worker code while using R2 storage. In the code I can see the file being cached and retrieved but I noticed there was no report of it in the Cache Dashboard for my domain. In addition there is no header for CF-Cache-Status that says hit for example or any status showing. Is this expected? Update: disregard i got it 😄

When we uses Services Bindings, someone answered me here that there is no extra cost ( 1 counted request on the worker X that calling worker Y ). So, what happen if the 1st worker is bundled but call a unbounded worker or a DO ? Is the cost is adjusted to 1 DO / unbound worker rather 1 worker, or is it count like 1 worker + 1 unbound/DO worker ?

I would imagine it counts as 1 bundled & 1 unbound

If this is the reality, this add complexity for developers and seems useless. Imagine I have a 1st worker bundled (B), that need to call an worker unbundled (U). If the pricing is based on 2 requests rather than one in this scenario because it's not the same kind of worker, this mean many developers will just call a 3rd unbundled (A) worker in front of them like this: (A) => (B) => (U), and this will be cost 1 request of unbound.

You'd still have the bundled worker initially? That's going to cost exactly the same

For me, CF should still count 1 request (still limited to the 50ms) but if the sub-request call a DO/unboundled worker, that should increase the cost to 1 request of an unbound, rather than 1 request of a bundled

One unbound worker calling another will be billed as one joint duration - that's nothing to do with bundled workers

are you sure? Because the duration of CPU time doesn't count the fetched subresources in therory