🦋 Changeset detected

Latest commit: 38859dc

The changes in this PR will be included in the next version bump.

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

npm i https://pkg.pr.new/create-cloudflare@11301

npm i https://pkg.pr.new/@cloudflare/kv-asset-handler@11301

npm i https://pkg.pr.new/miniflare@11301

npm i https://pkg.pr.new/@cloudflare/pages-shared@11301

npm i https://pkg.pr.new/@cloudflare/unenv-preset@11301

npm i https://pkg.pr.new/@cloudflare/vite-plugin@11301

npm i https://pkg.pr.new/@cloudflare/vitest-pool-workers@11301

npm i https://pkg.pr.new/@cloudflare/workers-editor-shared@11301

npm i https://pkg.pr.new/@cloudflare/workers-utils@11301

npm i https://pkg.pr.new/wrangler@11301

commit: b88b00f

Converting to draft since the C3 e2es don't currently run in CI and we should enable them before merging and new C3 autoconfig PR

When I tried this locally, with a new project, I got "✘ [ERROR] The entry-point file at ".open-next/worker.js" was not found.". I needed to do another npm run deploy manually

The issue is that it doesn't run opennextjs-cloudflare build. I'm not sure what the best workaround is here.

Should it change the build command?

Should it change the build command?

oh... yeah I see the issue.... that's annoying.... I don't think we could change the build command since that's for example used by the Vercel CLI and changing it can have some unexpected consequences... 🤔 I think we should ideally let wrangler deploy know the build command that it needs to run? 🤔

@ascorbic WDYT?

0927724

0927724

I think these changes should be in a separate PR. Will they then override the package.json build command too?

I’ll try to create a minimal reproduction later, but I think the open-next project detection might be too permissive. I have a Wrangler config at the repo root that deploys an Astro app. Then I have a workspace with a Next.js app that uses open-next. The issue is that when I deploy the root app, it gets identified as an open-next project, which is incorrect, and it results in this error:

See https://github.com/ariakit/ariakit/actions/runs/20287760629/job/58265344982

Do you have this documented somewhere? Because in my case, the new Wrangler version (4.55.0) is not respecting the deployment parameters, such as --env, --name and --dispatch-namespace.

As showing below, first it runs Wrangler deploy with some parameters, then it calls opennextjs-cloudflare deploy without any parameter.

Running Wrangler Commands
  /opt/actions-runner/_work/_tool/node/20.19.6/x64/bin/npx wrangler deploy --name projectcom-dev-merge-17-dec --env=nonprod --dispatch-namespace wfp
  Proxy environment variables detected. We'll use your proxy for fetch requests.
  
   ⛅️ wrangler 4.55.0
  ───────────────────
  OpenNext project detected, calling `opennextjs-cloudflare deploy`
  Proxy environment variables detected. We'll use your proxy for fetch requests.
  
  ┌──────────────────────────────┐
  │ OpenNext — Cloudflare deploy │
  └──────────────────────────────┘
  
  Incremental cache does not need populating
  Tag cache does not need populating
  Proxy environment variables detected. We'll use your proxy for fetch requests.
  
   ⛅️ wrangler 4.55.0
  ───────────────────
  ▲ [WARNING] Multiple environments are defined in the Wrangler configuration file, but no target environment was specified for the deploy command.
  
    To avoid unintentional changes to the wrong environment, it is recommended to explicitly specify the target environment using the `-e|--env` flag.
    If your intention is to use the top-level environment of your configuration simply pass an empty string to the flag to target such environment. For example `--env=""`.

This change just took down a production site because --env wasn't passed to the opennext deploy. When it failed, it assigned the prod custom domains to our staging env and stole them from the prod deploy. This happened completely isolated in the staging worker. The cloudflare infra just let a config value take the custom domains from our prod instance and apply them to the staging instance like it was no problem.

meanwhile the sites that used routes instead failed with an error saying the route was already in use

@dario-piotrowicz We deploy our apps to Workers for Platforms by setting --dispatch-namespace for wrangler deploy. With new version of wrangler it doesn't work anymore. How can I disable it?

@pliashkou you can temporarily pin your wrangler version to 4.54 to avoid this. We have made some improvements in 4.56 which should have resolved the OpenNext priority issue for detection, but if that's still broken for you pinning a previous version will get you around the problem whilst we investigate. For safety we have deprecated 4.55 and 4.56 until after the holidays.

To do this you can specifically target the version of wrangler via npx wrangler@version deploy so npx [email protected] deploy

@diegohaz @jkribeiro - see the above comment re: pinning to 4.54.

we have deprecated 4.55 and 4.56 due to this regression.