Forbid explicit step in slice

Rule request

Thesis

# bad
a[1:4:]
# good
a[1:4]

# bad
a[1::]
# good
a[1:]

# bad
a[1:4:1]
# good
a[1:4]

# bad
a[1::1]
# good
a[1:]

# bad
a[::1]
# good
a.copy()

Case with a[::] will be covered in #1011

Reasoning

Consistency

[skarzi]

So, why could we want to use [::] on it if it is immutable? Also, you can always use copy.copy function for custom collections (than does nothing for tuples as well, btw).

True, it doesn’t make much sense when we are talking about immutable objects, I just wanted to give some example that .copy() is not a good solution, however I agree that copy.copy() is.

As you said, we can’t say the difference between a[:] and a[::]. So we can’t suggest using one instead of others.

If we decide to write token visitor instead of AST, it will be possible to distinguish cases like 1st and 2nd ones, but I need to do some research and experiments to fully confirm that.

[orsinium]

I concluded that for 1st and 2nd case the AST looks identical

Let’s for the beginning cover remaining cases. It’s better than nothing.

tuple is immutable so it doesn’t have such method

So, why could we want to use [::] on it if it is immutable? Also, you can always use copy.copy function for custom collections (than does nothing for tuples as well, btw).

however we can still suggest to use a[:]

As you said, we can’t say the difference between a[:] and a[::]. So we can’t suggest using one instead of others.