app: e2e-tests: Make the desktop suite runnable as documented

[unlikelyzero]

Summary

Two problems stopped the desktop e2e suite from being runnable by following the documentation.

npm run app:test:e2e failed immediately with Missing script: "test". The root script delegated to a test script in app/e2e-tests that does not exist; that package only defines test-app and test-web.

Separately, the README omitted a required setup step. app/package.json declares "main": "build/main.js" and the specs launch the app with electron ., which resolves that field, but nothing in the README or the specs created it.

Related Issue

Fixes #6633
Fixes #6640

Changes

• Root app:test:e2e now points at test-app, the mode these tests are written for. Adding a delegating test script to app/e2e-tests would also have fixed the error, but it puts another npm run in the chain and arguments are not forwarded across those, so flags would be silently dropped.
• Added a pretest-app hook that runs compile-electron -- --dev, so the Electron entry point is built automatically and the suite cannot be run against a missing one. That seemed better than documenting one more manual step people can skip.
• Rewrote app/e2e-tests/README.md around what the suite actually requires.

The README changes are based on getting the suite running rather than on reading the source. The previously undocumented requirements:

• A minikube cluster named minikube. The name matters: clusterRename.spec.ts expects it, and clusterAutoConnect.spec.ts shells out to minikube directly to create and delete its own throwaway profile, so both minikube and kubectl must be on PATH.
• The frontend and backend built, because app mode runs the app from source. With ELECTRON_DEV=true, app/electron/main.ts resolves ../frontend/build/index.html (main.ts:91) and ../backend/headlamp-server (main.ts:772).
• Chromium installed, even though these are Electron tests. Every spec requests Playwright's page fixture before switching to the Electron window, and requesting it launches a browser. This one is easy to miss.

It also documents that flags cannot be passed through npm run app:test:e2e from the repository root, because that delegates to a second npm run; run the script from app/e2e-tests when you need flags.

Steps to Test

The script fix, which needs no cluster:

# Before: Missing script: "test"
# After: resolves and starts Playwright
node -p "require('./package.json').scripts['app:test:e2e']"
# -> cd app/e2e-tests && npm run test-app

The pre-hook, which is the point of the second fix:

rm -f app/build/main.js
cd app/e2e-tests && npm run pretest-app
ls -la ../build/main.js     # rebuilt

That the suite is discoverable, without running it:

cd app/e2e-tests
PLAYWRIGHT_TEST_MODE=app npx playwright test --list
# -> Total: 4 tests in 3 files

Then follow the README from a clean checkout and confirm you reach a running suite.

Screenshots (if applicable)

Not applicable, docs and npm scripts only.

Notes for the Reviewer

• This does not make the suite pass. It currently fails for unrelated reasons, tracked in Desktop e2e suite fails and modifies the developer's kubeconfig #6648 and fixed in app: e2e-tests: Fix the desktop suite and stop it mutating the kubeconfig #6641. This PR only makes it possible to invoke it by following the docs. I have kept the two separate because they are different problems with different reviewers in mind, and app: e2e-tests: Fix the desktop suite and stop it mutating the kubeconfig #6641 is test-code while this is scripts and documentation.
• On pretest-app versus documenting the step. npm runs pre<script> automatically for npm run test-app, so this costs nothing at the call site. The alternative was a README line saying "remember to run compile-electron first", which is the kind of step people skip and which produced a confusing failure. Happy to switch to documentation only if you would rather the scripts stay minimal.
• Web mode is unchanged. Only test-app gets the pre-hook, because web mode does not launch Electron and so does not need build/main.js.
• I did not touch the top-level e2e-tests README, which has a similar but separate mismatch tracked in e2e-tests/README.md setup does not match what the tests require #6634.

[Copilot]

Pull request overview

This PR fixes the developer entry point for running the desktop (Electron) Playwright e2e suite and updates the accompanying README so the suite can be invoked by following the documented steps.

Changes:

• Updated the root app:test:e2e script to call test-app (the script that actually exists in app/e2e-tests).
• Added a pretest-app hook in app/e2e-tests to ensure the Electron entry point (app/build/main.js) is compiled before tests run.
• Rewrote app/e2e-tests/README.md to document the real prerequisites and workflows for app vs web mode.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
package.json	Fixes the root e2e script to invoke the correct desktop test script (test-app).
app/e2e-tests/package.json	Adds pretest-app to build Electron dev output before running app-mode Playwright tests.
app/e2e-tests/README.md	Updates documentation to reflect actual prerequisites and how to run app/web modes.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

Comments suppressed due to low confidence (1)

app/e2e-tests/README.md:77

• This note says flags cannot be passed through npm run app:test:e2e, but they can be forwarded if you include an extra -- so the second npm run receives the arguments. As written, readers may conclude it’s impossible rather than just slightly awkward.

Note that flags cannot be passed through `npm run app:test:e2e` from the root,
because that delegates to a second `npm run` and the arguments are not
forwarded. Run the script from `app/e2e-tests` when you need flags.

[illume]

Thanks! 🎉

[kubernetes-prow]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: illume, unlikelyzero

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Details

Needs approval from an approver in each of these files:

• OWNERS [illume]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment