what's everyone working on today?

Just preparing small project for sharing in #783765338692386886 ! 😈

KV 😜

Hey everyone, it seems like the header limit on response headers in CF Workers is 4k per header. Is there any chance this could be increase to something like 16k? I have a use-case where I need to pack up to 16k into the response headers to include metadata. I can split this up right now into multiple headers, but that's a bit hacky

Another question: Is there a way to do cache purging based on a key? So basically "tagging" a cache item with a bunch of keys, with which I can then later purge it. Both Akamai and Fastly support that.

There is a way for Enterprise accounts https://support.cloudflare.com/hc/en-us/articles/200169246-Purging-cached-resources-from-Cloudflare#h_6d756ac9-c476-45e8-a5d4-e2a6e45d9dc7

Ok thanks. For now I'm not planning to get an enterprise account 😄

please stop

@User You basically can do that already if you use workers KV.

You can get the locale from the headers (or route, however you want) and pass that as a param to get the translated content.

If you want to store the strings as keys instead of the locale. Like:

"Hello": { ..., jp: "こんにちは", fr: "bonjour" }

It is still good to use some sort of template language like in the doc above because you can get multiple rewrites or partial rewrites without one.

how did you get here without knowing what cloudflare is @User

I have a question. Is that discord some kind of support or only to promote cloudflare?

Kinda support but specifically for Workers. Like a real-time version of the community forum.

So in which channel i can ask if i have a question?

hello, welcome :). You can ask a question in any channel, however, this one is the best if you just have a general Workers question

Ok. Maybe its not the worker only but i want to ask something. I played a bit with some configuration options and now im stuck with a problem. I have an AWS S3 bucket that i want to reach with a worker. In the worker i can get an object from the bucket and get a status 200 response with the image, i made a request for. But when i try to use that worker with a route and a path of my domain and get an access denied repsonse. My worker looks like this.

javascript
addEventListener('fetch', event => {
  event.respondWith(handleRequest(event.request));
})

async function handleRequest(request) {
  let requestUrl = new URL(request.url).pathname;
  requestUrl = requestUrl.slice(4, requestUrl.length);
  const url = "https://UrlToProvider/bucketName.de"
      + requestUrl;
  return await fetch(url);
}

My route is

https://my-domain.com/img/*

When i now go to the browser and use the url

https://my-domain.com/img/path/to/img

I normally should get that image, but something strange happens. I get an Access Denied reponse and in this response he searched for a key in the bucket. From this example he searches for

<key>/path/to/img</key>

But thats the wrong key. The slash at the start is to much. But my url still looks fine. I tried to play a bit with the slice function for the requestUrl, but that only breaks everything. As i mentioned, when i pass a real url to the worker, the worker can get the image and recieve a status 200 from the request.

My key is as example path/to/img but he searches for /path/to/img. But only outside of the worker with the route. Inside of the worker everything works fine.

hmm did you enable public read access for your S3 objects?

Yes, i can access the bucket over the original url from the provider

I just dont know why he adds this slash before the key. Im confused.

ohh ehm. I changed something with the access and now it works. Was still using the IPv4 and IPv6 list of cloudflare in the bucket policy. I removed the condition in the bucket for these ips. But why did that worked in the worker itself before?

were you using workers preview?

The question from @User stands, but notice that when going to the origin, maybe S3 is reading the proxied address so it knows it's coming from Cloudflare but from you?

Yes i use the quickedit on the cloudflare site. Do you mean that?

quick edit on the cloudflare site is run on GCP

so the outbound requests will be from GCP ips, not cloudflare ips

So i should use the wrangler stuff i found in the docs? I mean would you recommend it?