refactor: migrate tests from tape to node:test

[Tony133]

Proposal:

This PR migrates the test suite from tape to the native node:test runner and node:assert module. Since tape is no longer actively maintained and Node.js now provides a robust built-in testing solution, this migration reduces external dependencies and align the project with other repositories in the fastify ecosystem

Changes made:

• Migrate tests from tape to node:test
• Removed airtap, airtap-playwright with node:test along with playwright

[metcoder95]

rest lgtm, just question on removing the browser tests

[metcoder95]

Shall we remove it?

[Tony133]

Shall we remove it?

I removed the section related to test-browser from ci.yml, so I think it should be okay now we should then see if it is aligned with the others fastify-libraries

[gurgunday]

#178 (comment)

Lgtm other than this

[climba03003]

Browser test exists for reason, it allows us to check whether the code break in browser or not.
For example, Buffer check inside the codebase is not compatible in browser.

[Tony133]

Browser test exists for reason, it allows us to check whether the code break in browser or not. For example, Buffer check inside the codebase is not compatible in browser.

Okay, thank you for the clarification! so we have to leave them as they are: does that mean we can't replace tape with node:test? 🤔 Maybe I could replace airtap, which is now obsolete, with node:test along with playwright

[Tony133]

I thinking about it, we can still replace tape with node:test. Since airtap is no longer actively maintained and doesn’t work with the Node.js test runner, my proposal is to migrate browser testing so that we use node:test directly alongside Playwright. This allows us to continue testing the code in a real browser environment without having to rely on outdated tools.

Anyway, I made a commit and put the browser tests back into CI , see commit here: fe282a8

[climba03003]

It doesn't seems to use playwright for testing?
Only installing playwright do not means it is inside the browser environment.
See https://playwright.dev/docs/ci#github-actions

[Tony133]

Okay, I think I've fixed it. I added a file named browser.spec.js ( see commit here: 846caba) to run the test in the browser. Using Playwright in automation script mode (via browser.spec.js) allows us to test the real environment in a more direct, clean way, without any “magic” behind the scenes like airtap ( before that, Airtap read everything from the index.js file) , thereby filling the gap in cross-browser testing. Without this script, we would no longer have visibility into the library’s behavior within the JavaScript engines of various browsers (Chromium, Firefox, WebKit).

The new test does not merely re-run the unit tests (which remain in Node), but specifically focuses on the two critical issues of the client-side environment: the absence of Node-only objects (such as Buffer) and the validation of the prototype pollution protection logic within the browser's JavaScript engine.

[Fdawgs]

Away from a computer atm and reviewing from phone is hard, but any reason to use this over the testcontext assert inside of test functions?

[Tony133]

Away from a computer atm and reviewing from phone is hard, but any reason to use this over the testcontext assert inside of test functions?

Yes, you can do that with TestContext as well ( see commit here: cb48c17) ; you can do it both ways 😉