The recent commits in the Workers SDK are hinting at Workers supporting assets like Pages does, wher

The recent commits in the Workers SDK are hinting at Workers supporting assets like Pages does, where hopefully there will be a possibility that you won't be charged workers invocations for assets, which will be awesome for e.g. running next-on-pages from workers instead of pages, getting e.g. support for tail worker there, August 15th?

https://github.com/cloudflare/workers-sdk/pull/6325

IItsWendell The recent commits in the Workers SDK are hinting at Workers supporting assets l...

Hard@Work•8/7/24, 12:25 PM

This is the building blocks for Convergence, i.e., making Workers and Pages a single product

Zack•8/7/24, 12:30 PM

Just want to get some eyes on this if possible, maybe better documentation is needed but this killed quite a few hours this morning and seems extremely poorly documented at best, and dangerous at worst as it could silently be unencrypting traffic without people realising. workers-and-pages-helpFetch getting infinite 308 redirects in worker but fine locally

Zack•8/7/24, 12:30 PM

This is not a request for support anymore, we figured out the problem, but it feels extremely hidden that this is the behaviour, even knowing about the normal "orange cloud" proxy behaviour with these settings

Zack•8/7/24, 12:32 PM

I do wonder how many people are unknowingly doing this, with auth tokens and other sensitive data just flying around in plaintext between cloudflare and their origin

HHard@Work This is the building blocks for Convergence, i.e., making Workers and Pages a si...

ItsWendellOP•8/7/24, 12:46 PM

I'm excited

Mmarsoleg Hi, my program needs to do caching to work. I read upstash documentation and lik...

bikevit2008•8/7/24, 1:05 PM

https://discord.com/channels/807028371451936838/1267792517537992786/1267819161933971549

EErisa upstash support for workers has been around for much longer than tcp support has...

bikevit2008•8/7/24, 1:05 PM

https://discord.com/channels/807028371451936838/1267792517537992786/1267819161933971549

bikevit2008•8/7/24, 1:05 PM

I did tests with workers tcp connections to upstash

bikevit2008•8/7/24, 1:05 PM

raw tcp

bikevit2008•8/7/24, 1:05 PM

Latency more stable and faster

bikevit2008•8/7/24, 1:06 PM

bikevit2008•8/7/24, 1:07 PM

I can share my worst code did in night hours for you, i did it with gpt4 help, but it works good, but i don't sure what him give powerful coverage all opportunities with upstash or redis

bikevit2008•8/7/24, 1:11 PM

But you can extend my tech-MVP

Isak•8/7/24, 2:57 PM

Anyone who works with smart placement? workers-and-pages-discussions

henri•8/7/24, 2:59 PM

Does anyone know how to add the Workers API to ESLint globals? I'm trying to configure ESLint to recognize the Workers API objects without throwing undefined errors.

iska•8/7/24, 5:14 PM

is it a terrible bad idea to dynamically import a heavy lib say 4mb where we are caching it after first initial load in a cdn for faster delivery, at runtime to make use of it inside a worker ?

Zane•8/8/24, 6:53 AM

I'm trying to add a watermark to an image using Cloudflare Workers and the image transformation feature. I've followed the documentation, but the watermark doesn't appear on the resulting image. Here's my code for the watermark function. Could someone help me identify what I'm doing wrong or if there are any additional steps I need to take to make this work?

const addWatermark = async (c: any, imageUrl: string, watermarkUrl: string) => {
try {
const response = await fetch(imageUrl, {
cf: {
image: {
draw: [
{
url: watermarkUrl,
bottom: 10,
right: 10,
fit: 'contain',
width: 100,
height: 50,
opacity: 0.8,
},
],
},
},
});

if (!response.ok) {
throw new Error(Failed to add watermark: ${response.status} ${response.statusText}Failed to add watermark: ${response.status} ${response.statusText});
}

const blob = await response.blob();
const key = watermarked-${Date.now()}.webpwatermarked-${Date.now()}.webp;
await c.env.MY_R2_BUCKET.put(key, blob);

return ${c.env.R2_PUBLIC_URL}/${key}${c.env.R2_PUBLIC_URL}/${key};
} catch (error) {
console.error("Error adding watermark:", error);
throw error;
}
};

Zane•8/8/24, 6:53 AM

i use hono app

ItsWendellOP•8/8/24, 11:15 AM

Service bindings should be close to 0ms latency right? I've a worker and a pages project which are connected through service bindings (RPC). The worker has 2 RPC functions, setEnvironmentVariablesetEnvironmentVariable and getEnvironmentVariablegetEnvironmentVariable. The pages project calls getEnvironmentVariablegetEnvironmentVariable, but when I track the time it takes to get a value (line before the RPC call and after (startMs, durationMs), it's more often 100-300ms than it is 0-30ms (it does hit it, but very rarely). Any reason why? I thought that service bindings would bind in the same isolate.

The implementation of getEnvironmentVariable is essentially: return this.env[key]return this.env[key], setEnvironmentVariablesetEnvironmentVariable uses the Cloudflare API to set an environment variable on that worker.

kian•8/8/24, 11:22 AM

Are you doing this getEnvironmentVariable at the start of your handler?

ItsWendellOP•8/8/24, 11:31 AM

No, it's in SSR on next-on-pages that I call the RPC function getEnvironmentVariable, but I'm comparing Date.now() right before and right after the RPC call, which I then log into a log aggregator, and the implementation of the RPC function in the WorkerEntrypoint of that worker is:

/** Edge Config Worker */
export default class extends WorkerEntrypoint<EdgeConfigEnv> {
  async setEnvironmentVariable(name: string, value: string): Promise<boolean> {
     const res = await fetch(..., {}); // Simple POST call to `/workers/scripts/${scriptName}/secrets`
  }

  getEnvironmentVariable(name: string): string | null {
    if (typeof this.env[name] === "string") {
      return this.env[name];
    }

    return null;
  }
}

/** Edge Config Worker */
export default class extends WorkerEntrypoint<EdgeConfigEnv> {
  async setEnvironmentVariable(name: string, value: string): Promise<boolean> {
     const res = await fetch(..., {}); // Simple POST call to `/workers/scripts/${scriptName}/secrets`
  }

  getEnvironmentVariable(name: string): string | null {
    if (typeof this.env[name] === "string") {
      return this.env[name];
    }

    return null;
  }
}

IItsWendell No, it's in SSR on next-on-pages that I call the RPC function getEnvironmentVari...

ItsWendellOP•8/8/24, 12:08 PM

@kian This has been for a while like this btw, I could share the history of logs around this topic if interested!

iska•8/8/24, 1:25 PM

given each worker is creating a .wrangler that has local state like d1

imagine i have multiple workers and i want them to use the same local db

is there a way to do so instead of each worker using its own isolated local state ?

iska•8/8/24, 1:25 PM

again i m only referring to local state with different workers using same db, in prod this is not an issue because obviously they are pointing to the same db

Original message was deleted

Chaika•8/8/24, 1:40 PM

I don't believe that's true https://www.sqlite.org/draft/faq.html#q5, SQlite's pretty awesome, if you have WAL enabled you can avoid entire db locks too

iska•8/8/24, 1:40 PM

huh ? it s not true i mean in prod this is standard multiple workers writing to the same db

Iiska huh ? it s not true i mean in prod this is standard multiple workers writing to ...

Chaika•8/8/24, 1:41 PM

way different in prod vs local

Chaika•8/8/24, 1:41 PM

Prod your D1 DB is a Single Threaded Durable Object, in dev it's just sqlite and an sqlite file

Chaika•8/8/24, 1:43 PM

You can specify

--persist-to

--persist-to

in the dev command https://developers.cloudflare.com/workers/testing/local-development/#work-with-local-data, https://developers.cloudflare.com/workers/wrangler/commands/#dev to specify a specific folder to persist stuff too, while I think sqlite itself would be fine idk if the rest of it would if you're talking about running it at the same time

detectiveConan•8/8/24, 1:44 PM

hey @Chaika dude can you help me out as I am trying to add security headers to my website. but I tried to add them via nagivating to rules>transform rules>modify response header and adding it but still not reflecting on checked by securityheaders?

Chaika•8/8/24, 1:44 PM

doesn't sound related to workers, make a #general-help thread including the rule you made and your website url?

CChaika You can specify `--persist-to` in the dev command <https://developers.cloudflare...

iska•8/8/24, 1:46 PM

yes but the thing is by default each worker create its own local state .wrangler where the sql file is set, so each worker pointing to another db , and I would want them to share it , is there a way to do so ?

CChaika doesn't sound related to workers, make a #general-help thread including the rule...

detectiveConan•8/8/24, 1:46 PM

even I tried with deploying workers but that also not help. shall I share my script here and you can check wether it's correct or not?

iska•8/8/24, 1:47 PM

Nevermind

DdetectiveConan even I tried with deploying workers but that also not help. shall I share my sc...

Chaika•8/8/24, 1:47 PM

you don't need a worker, would be a waste of requests/paying if you just want to add headers, stick with transform rules

Iiska yes but the thing is by default each worker create its own local state .wrangler...

Chaika•8/8/24, 1:47 PM

the --persist-to flag can be used to override where that local state folder is

iska•8/8/24, 1:47 PM

Persist to was what I was looking for

iska•8/8/24, 1:47 PM

Exactly

iska•8/8/24, 1:47 PM

Thanks

iska•8/8/24, 1:47 PM

do I have to do something to avoid lock issue using that ?

ArmoredCavalry•8/8/24, 3:13 PM

Not sure if ok to cross-post my help request here, but if so - 'Global' Worker for Logging (alternative for logflare App)

delta•8/8/24, 6:21 PM

Is it common for worker to take some time (e.g., 1 second for a 100 MiB transfer) to complete a PUT operation to the R2 storage after the user has sent the entire request body?

rawwerks•8/8/24, 6:38 PM

I'm following along here : https://developers.cloudflare.com/workers-ai/configuration/open-ai-compatibility/, with one question:
do i need to set node_compat = truenode_compat = true (in my

wrangler.toml

wrangler.toml

) in order to use the openai node/TS SDK? (import OpenAI from "openai";import OpenAI from "openai";)

iska•8/8/24, 7:42 PM

someone could share me a good flow of production grade ci/cd for workers apps and services ?

will take anything example, blog, tuto

like what's the idea ?

run tests
deploy in preview
then add on demand deploy to prod
?

not sure what should i do more for workers usually with regular services you have lot of steps involving bundling and optimizations like docker, kubernetes yada yada but this part seems to be taken care of automatically