Change behavior of String.right

[KoBeWi]

Follow-up to #28648

Before:

print("Hello world".right(3)) #lo world

After:

print("Hello world".right(3)) #rld

Basically opposite of left(), so is more intuitive and consistent.

I replaced all source occurrences of right to substr(), which has the same behavior as old right. btw, right was calling substr anyways, so every instance that used it might be few µs faster now 🤔 (heh)

[groud]

It would be good to implement, with both left() and right(), a behavior with negative indices.
So that print("Hello world".right(-3)) Would print lo world. (Basically this avoid typing print("Hello world".right(len("Hello world")-3))

[KoBeWi]

^
I added this as another commit (for now ofc).

EDIT:
Also mentioned it in docs.

[KoBeWi]

I have problems rebasing this due to renamed/moved files. Any advice welcome >_>

[madmiraal]

I've had a look at trying to rebase your PR and it's going to be difficult. There are over a year's worth of commits that, as you identify, include several files that have been renamed and others that have been completely refactored. Considering the number of lines being changed in this PR vs the number of conflicts, it may be easier to recreate it, or I can reopen #47434.

[KoBeWi]

Ok rebased. I manually copied my string changes and then modified other files from scratch. Also updated some tests.

[groud]

I accepted the change, but I think it might have been good to keep calls to right instead of substr whenever possible. Likely by using the new behavior with negative indices.

[KoBeWi]

Why though? Right() serves a different purpose now, messing with indicies will only cause confusion.

[groud]

Because right is a lot more explicit compared to substr regarding what it does. substr, from it's name, does not indicate if you take the right or the left part of the string.

It's not a big deal anyway but IMHO, it would make the code easier to understand.

[akien-mga]

I accepted the change, but I think it might have been good to keep calls to right instead of substr whenever possible. Likely by using the new behavior with negative indices.

I wonder about this too, but IMO there's a big semantic shift with this PR and the calls to right before the PR meant something very different semantically to what right would convey from now on.

Before the PR, we'd do things like str.right(1) to express "drop the first character, give me the rest", which doesn't actually have much to do with the right part of the string - it's more about how many characters to cut from the left.

After this PR, to convey the same meaning, str.substr(1, -1) might be a better way to convey this "cut one char, give the rest" behavior. If you want to keep using right(), it would have to be str.right(str.length() - 1) i.e. "give me a string with all characters minus one starting from the right".

If what you suggested was to use str.right(-1), it would not give the same result as what str.right(1) did before the PR.

[akien-mga]

Needs a rebase, otherwise I think it's good to go.

[KoBeWi]

Rebased again. And added some tests again (for negative arguments).

[akien-mga]

Thanks!