[feat] a11y-no-redundant-roles

[jamesb93]

This solves one task on the list in #820. My first PR so probably a lot going on wrong here but I would be very appreciative for feedback and guidance on contributing.

Before submitting the PR, please make sure you do the following

• It's really useful if your PR references an issue where it is discussed ahead of time. In many cases, features are absent for a reason. For large changes, please create an RFC: https://github.com/sveltejs/rfcs
• Prefix your PR title with [feat], [fix], [chore], or [docs].
• This message body should clearly illustrate what problems it solves.
• Ideally, include a test that fails without this PR but passes with it.

Tests

• Run the tests with npm test and lint the project with npm run lint

[jamesb93]

Thanks for your feedback @bluwy. Youre suggestions and enhancements have been implemented in db606c7, 91ec3fc, 53b8919, 17906c2.

[dummdidumm]

I just realized there exists a PR already for this: #5361 . Though I'm not sure which implementation should be preferred.

[bluwy]

Crap now I feel bad for choosing either one since both has its good parts and effort put into it. #5361 does have more checks. On the other hand, this PR also handles parent section/article checks. If I have to choose, I think we should continue this, and borrow over the extra checks from #5361 to here. We can then have @mhatvan as the co-author in the commit description.

[jamesb93]

Crap now I feel bad for choosing either one since both has its good parts and effort put into it. #5361 does have more checks. On the other hand, this PR also handles parent section/article checks. If I have to choose, I think we should continue this, and borrow over the extra checks from #5361 to here. We can then have @mhatvan as the co-author in the commit description.

I'd be more than happy to update with work already done in #5361. I think the joining of the two would make for a robust set of checks.

[jamesb93]

I've updated this PR with the conditions that are checked in PR #5361 @bluwy + @dummdidumm. I've also updated the logic so that there isn't a massive if statement with numerous nested ||. Instead there is now a Map() that contains the many cases that can arise with anything else being caught by this.name === value.

[bluwy]

Thanks for further improving the PR! I've left some comments to help with performance, but after that this looks good to me.

[bluwy]

Thanks for keeping up with the many requested changes 😅 LGTM with one tiny nitpick.

[mcmxcdev]

Well, I tried to get attention to my PR without success for a while. Anyway, I am fine with having this PR merged with stuff taken from mine ;)
The most important part is to get more a11y checks into svelte!

[jamesb93]

Well, I tried to get attention to my PR without success for a while. Anyway, I am fine with having this PR merged with stuff taken from mine ;) The most important part is to get more a11y checks into svelte!

It is definitely important to acknowledge that a significant portion of the code was championed by you @mhatvan. I learned from it and it looks like the checks should be pretty robust as a result. Thank you :)

[mcmxcdev]

Thanks man, I appreciate it!

All that matters is more and improved a11y checks ;)

[jamesb93]

It seems like one specific test (unrelated perhaps to this PR) is failing on macOS-latest and node 12. Is there any guidance on making it so this PR is fully test compliant?

[dummdidumm]

Don't worry about that, according to CI they just timed out.

[Conduitry]

Released in 3.45.0 - thank you!

[jamesb93]

Released in 3.45.0 - thank you!

Woo! Thanks everyone for your guidance and hard work. Much appreciated and I learned a lot.