[pickers] Fix keyboard navigation with yearsOrder="desc" and direction="rtl"

[thomasmoon]

Hi @LukasTy @MBilalShafi ,

A thousand apologies, I had tested this during development but when yearsOrder was changed from a boolean to asc | desc the RTL horizontal direction check got borken.

Still works fine with the default settings but if yearsOrder='desc' is used with rtl now the left/right navigation is backwards when the theme is right to left.

Finally found a good place to test these under the localization docs page and this small change should make everything right (/x/react-date-pickers/adapters-locale/).

I will work on an update to the automated tests too that would catch this, but this quick fix for now.

• I have followed (at least) the PR section of the contributing guide.

[mui-bot]

Deploy preview: https://deploy-preview-14682--material-ui-x.netlify.app/

Generated by 🚫 dangerJS against 9317740

[flaviendelangle]

Thanks for the quick fix !

I will work on an update to the automated tests too that would catch this, but this quick fix for now.

The next release is next Thursday, so we should be able to have 1-2 tests by then 👍

[LukasTy]

Thank you for the super quick fix! 🙏
I'm sorry that I did not spot this before approving. 🙈

As Flavien has said, by default, we would release at the end of next week.
This does not look like a critical bug and can probably wait for tests if you choose to add them. 👌

[thomasmoon]

Thank you for the super quick fix! 🙏 I'm sorry that I did not spot this before approving. 🙈

As Flavien has said, by default, we would release at the end of next week. This does not look like a critical bug and can probably wait for tests if you choose to add them. 👌

I got the coverage added. The tests may not be the most beautiful, but the two cases of left and right in RTL mode definitely break in the previous version, so this could do the job! There weren't that many keyboard nav tests in the project to take a model from but this is surely a good addition. Happy to hear any feedback about the approach.

One interesting thing I ran into was that an additional Enter keyboard press would not trigger the onChange as expected, but it wasn't actually needed here because the focus change can be tested just by checking the active element. Now that I said that I realize I can remove that event from the tests.

[flaviendelangle]

Looks good to me
Thanks for taking care of it!