Style/AsciiComments default Enabled is true

[esotericpig]

I know that you can change it to be disabled, but it seems a bit odd that Style/AsciiComments is enabled by default. I only noticed when working on a Japanese project, and I put Japanese in a comment.

I understand wanting this for identifiers, constants, etc., but why for comments?

If it's to enforce English comments, I don't think it should be on by default. It sends the message that RuboCop is, by default, for English speakers and not for the global audience.

If it's to enforce Ascii encoding for the file, that is rare nowadays (since Ruby v2), so it seems like AsciiComments should still be defaulted to disabled (since the majority of Rubyists now have source files defaulted to UTF-8).

Thanks for the consideration, and I understand if it will remain defaulted to enabled.

[marcandre]

I agree. This also precludes emojis in comments.

[dvandersluis]

#72 for historical context.

This cop doesn’t really enforce English (if we even should), but just languages that use the Latin alphabet. I think it’s fine to have the cop for those who want it but even an English codebase can have a need for comments that contain non ascii.

Disabled by default makes sense to me.

[bbatsov]

Yeah, the idea was mostly to make sure that people stick to English in comments, with the assumption they'd likely have an international team working on the codebase. The cop was inspired partly from my own experience working in German/Austrian company where many people didn't know English yet the German and Austrian devs would often use German for identifiers and comments making it hard for people like me to understand what they meant. :-) OSS libraries are commonly documented in English as well.

I'm fine with retiring this cop, as agree it's somewhat controversial, plus detecting what's really English is pretty complex. Still, this can't happen before RuboCop 2.0. I guess we should also change the cop's name and improve its documentation.

[esotericpig]

I guess in a somewhat related note the enabled-by-default Style/NumericLiterals requires numbers to be grouped by 3s by default:

100_000

But Japanese (and probably Chinese as well?), like to group it in 4s:

10_0000

If want to make it more international-friendly, maybe either turn this off by default or add a groupby option to the cop.

Actually, I ran into this when testing with Chinese phone numbers, which are grouped by 4 in mainland China:

phone = 12_3456_7890

[JoshuaKGoldberg]

👋 coming from #5310 I'm interested in sending a PR. Would the team consider accepting one?

[dvandersluis]

@JoshuaKGoldberg how is this cop useful for you, aside from wanting it to accept emoji? I still think disabling it by default as discussed above makes the most sense.

We recently ended up disabling it in my work repo because it has too many edge cases (for instance, a comment containing a link with a non-ascii character, or a comment giving an example of what some code will generate which includes utf-8).

[JoshuaKGoldberg]

Exactly, I think disabling it by default makes sense.

[bbatsov]

Agreed.

[dvandersluis]

@bbatsov are we good with doing that now? I'm going to just do it now 😁

[bbatsov]

I think yes - it's the kind of cop that no one's going to notice got disabled.

[JoshuaKGoldberg]

Glorious, thanks!