Feedback on lowercase `\U`, `\u`, `\x` escapes introduced in 23.1 preview style #2916

[yilei]

I'm not sure if lowercase is the better decision, so I'm filing this feedback for discussion.

Reasons to favor lowercase:

1. repr() uses lowercases:

   >>> "\x1B"
   '\x1b'
   >>> "\u200B"
   '\u200b'
   >>> "\U0001F977"
   '\U0001f977'

2. I sampled 500K files from our codebase, lowercase is bit more popular (the count below is the total number escape sequences in strings):

   \x escapes:

   • lowercase: 440K
   • uppercase: 351K
   • no letters: 341K

   \u escapes:

   • lowercase: 144K
   • uppercase: 49K
   • no letters: 46K

   (This might actually be a reason to not have a preference, since this is polarized.)

Reasons to favor uppercase:

1. For \x escapes, Black already uses uppercase for hex numbers: 0xEF, so this will be consistent with Black itself.
2. For unicodes, https://unicode.org/ uses uppercases.

I personally think the uppercase reasons are stronger.

[JelleZijlstra]

For reference #2916 and #2067 is where this was added. I don't have a strong opinion here but this issue will provide a good way to gather user feedback while this is in the preview style.

[JelleZijlstra]

Thinking about this more and re-reading the original issue, I agree that we should use lowercase for these. I propose that we keep this change out of the 2024 stable style and make a preview change to lowercase.

[JelleZijlstra]

I think I misread something here, we already use lowercase in preview style, though @yilei above weakly prefers uppercase. I think the reasons for lowercase are a little stronger, and this has been in preview for almost the whole year with no complaints, so I'm inclined to stick with lowercase.

[yilei]

Feel free to close, as we haven't heard other feedback about this change.