Unable to build package in a local environment

[aegroto]

It would be beneficial to be able to build the package in a local environment. For instance, PRs such as #1606 require to verify that the bundle size is not affected too much. However, running the start npm run build command fails in the current project structure. I am currently using this Dockerfile:

# syntax=docker/dockerfile:1

FROM node:18.20.4-bullseye

ENV NODE_ENV=production

WORKDIR /app

EXPOSE 3000

COPY . .
RUN npm ci --legacy-peer-deps --loglevel verbose
RUN npm run build --verbose

The command docker build -t sn-test --no-cache . fails with a bunch of the following errors:

unhandledRejection TypeError [ERR_INVALID_URL]: Invalid URL
    at new NodeError (node:internal/errors:405:5)
    at new URL (node:internal/url:676:13)
    at /app/.next/server/chunks/4926.js:1:43211
    at 69496 (/app/.next/server/chunks/4926.js:1:43728)
    at t (/app/.next/server/webpack-runtime.js:1:127)
    at /app/.next/server/chunks/4926.js:1015:5058
    at t.a (/app/.next/server/webpack-runtime.js:1:866)
    at 95639 (/app/.next/server/chunks/4926.js:1015:4951)
    at t (/app/.next/server/webpack-runtime.js:1:127)
    at /app/.next/server/chunks/4926.js:3986:461 {
  type: 'TypeError',
  input: 'undefined',
  code: 'ERR_INVALID_URL'
}

I have uploaded the full logs and the update Dockerfiles in https://github.com/aegroto/stacker.news/tree/local-bundle-build.

I have been looking for a while on the web and it looks like these errors may be due to missing or wrongly formatted environment variables pointing to URLs. However I have manually checked and the .env files are copied in the resulting image, therefore the environment should be set up correctly. Is there any other step missing? Thanks in advance.

[huumn]

Did you copy .env.development -> .env.production? When NODE_ENV = 'production', it only uses .env.production.

[aegroto]

Thanks for your quick feedback. Do you mean that you actually need the variabiles in both .env.development and .env.production to be set? I thought only of the envs is supposed to be activated.

[huumn]

.env.production does not have all the production env variables in it because many of them are secrets

[aegroto]

I have been able to build by creating a .env file with variables from both development and production, adding VAPID_ variables as indicated here with a subject such as mailto:[email protected], then using this Dockerfile:

# syntax=docker/dockerfile:1

FROM node:18.20.4-bullseye

WORKDIR /app

EXPOSE 3000

COPY . .
RUN npm ci --legacy-peer-deps --loglevel verbose
RUN npm run build --verbose
CMD ["/bin/bash"]

There is no need to set NODE_ENV. the .env file is loaded automatically.

I will work on converting this to work in a GitHub action such that we can perform checks on the bundle directly during the CI pipeline.

[huumn]

Awesome!