PiniaColadaDevtools kills Cloudflare workers deployment completely

[Perdolique]

During the migration of my Nuxt.js app from Cloudflare Pages to Cloudflare Workers, I've found a bug that every second deployment completely kills the worker w/o the possibility to restore it — any following deployments don't help, only recreating the worker from scratch helped (error 1101).

Trying different variants, I've found that commenting following code helps. Uncommented code breaks the worker again. Tried a few times — commenting always helps.

<template>
  <!-- ... -->

  <!-- <PiniaColadaDevtools /> -->
</template>

<script setup lang="ts">
  // import { PiniaColadaDevtools } from '@pinia/colada-devtools'

  // ...
</script>

I don’t have a clean repro, since I’m already in the process of removing Pinia Colada from the project — I’ve realised the library is still pretty undercooked for real-world usage. Still, I filed the issue in case it helps.

I've used nuxt build and wrangler deploy scripts for testing and the Nitro preset cloudflare-module.

[posva]

I would contact cloudflare support. It might be bug there too. I can’t do much with this 😅
Devtools are stripped off in production

undecooked? What do you mean, this is not a dish 😂

[Perdolique]

Devtools are stripped off in production

Definitely not in this case. Uncommenting two strings above adds around 300 kB of overhead to the production build:

Σ Total size: 1.48 MB (469 kB gzip)
Σ Total size: 1.74 MB (558 kB gzip)

I already contacted Cloudflare, but the unrecoverable worker state issue on their side isn't related to the library.

[posva]

Are you building in production mode?

edit: I think the introduction of the disabled prop broke it. As a workaround, use <PiniaColadaDevtools v-if="isDev" /> and declare const isDev = process.env.NODE_ENV !== 'production'