lib/https: convert to using internal/errors

[ramimoshe]

covert lib/https.js over to using lib/internal/errors.js

ref: #11273

[lpinca]

Tiny nit: "a URL".

[ramimoshe]

Hi, who is responsible to merge this pr?

[lpinca]

@ramimoshe any collaborator can land a PR but we wait for 48+ h to make sure everyone had a chance to look at it.

[BridgeAR]

Is this code currently not covered? In that case it would be great to add a new test.

[BridgeAR]

CI https://ci.nodejs.org/job/node-test-pull-request/10306/

[BridgeAR]

Please use the already existing error type ERR_INVALID_DOMAIN_NAME instead of adding a redundant new one.

[BridgeAR]

LGTM even though I would strongly prefer to have a test added for this case before landing.

[BridgeAR]

@ramimoshe would you be so kind and add a test for this as well? That would be really great. If you need any help, please let me know.

[indutny]

LGTM

[Trott]

@BridgeAR This code is indeed covered in tests. test/parallel/test-http-invalid-urls.js: will exercise it. Unfortunately, that test appears buggy. It makes the perfectly reasonable but incorrect assumption that if the second argument in assert.throws() is a string, then it corresponds to the expected message. Alas, no. (This gotcha is documented.)

IMO, this PR should fix that test. This line:

const error = 'Unable to determine the domain name';

...should be changed to something like this:

const error = common.expectsError({code: 'ERR_INVALID_DOMAIN_NAME'});

(Maybe include the message property too? Maybe change the name from error to something like expectedError? But those improvements can happen subsequently, if at all.)

[Trott]

(Also, this will need a CI run before landing, of course.)

[Trott]

(Actually, I'm inclined to CI this now, land it, and then PR the test fix myself. If @ramimoshe wants to learn all about how we write our tests and stuff, awesome. But I'm going to avoid piling on changes for a first-time contribution.)

[Trott]

CI: https://ci.nodejs.org/job/node-test-pull-request/10328/

[Trott]

Hmmm....merge conflicts even though GitHub doesn't seem to notice them.

Squashed, rebased, force-pushed. Trying again.

CI: https://ci.nodejs.org/job/node-test-pull-request/10329/

[Trott]

Landed in 4843c2f

[Trott]

Welcome @ramimoshe and thanks for the contribution! 🎉

[ramimoshe]

Thanks everyone :)
Why did you close the PR without "merge"?

[ramimoshe]

BTW I will try to add tests in my other open PR

[lpinca]

@ramimoshe that's how we land PRs you can find more info on the process in our documentation.
@Trott already opened a PR for the test. See #15678.