Editorial: Clarify RegExp grammar parameter U → UnicodeMode

[mathiasbynens]

This makes the parameter more easily searchable while clarifying its meaning.

This came up while working on the draft spec for the RegExp set notation proposal, which introduces another RegExp grammar parameter that is related to Unicode. Rather than sticking to single letters, it seems nicer to move to more descriptive names.

If the editors are open to this change, I’d be happy to work on a patch that renames N as well (perhaps to WithNamedCaptureGroups).

cc @markusicu @markusicu @sffc

[markusicu]

Bike-shedding opinion:

• Longer grammar parameters would be more searchable (as you say) and more guessable, e.g., U vs. the Unicode boolean in the semantics
• We could use the same terms for grammar parameters and semantics booleans (e.g., Unicode), but it might also be confusing.
• For the syntax grammar, very long parameter names could be tedious to read & write, especially where multiple parameters are used on a line.
• Maybe intermediate abbreviations for the grammar parameters? E.g., U→Uni, (proposed) V→UniSet, N→Named

[mathiasbynens]

• Longer grammar parameters would be more searchable (as you say) and more guessable, e.g., U vs. the Unicode boolean in the semantics

+1

• We could use the same terms for grammar parameters and semantics booleans (e.g., Unicode), but it might also be confusing.

I think it would mostly be confusing, as it would imply a connection that isn't there.

• For the syntax grammar, very long parameter names could be tedious to read & write, especially where multiple parameters are used on a line.
• Maybe intermediate abbreviations for the grammar parameters? E.g., U→Uni, (proposed) V→UniSet, N→Named

That reduces searchability. Ctrl+F "Uni" would return lots of unrelated hits.

[jmdyck]

Looks fine now.

#2411 (review)

[michaelficarra]

I'd rather see us have a consistent editorial direction here, either going the verbose route for all grammar flags or the terse route for all grammar flags. We can discuss the tradeoffs of each in the editor call, but right now I don't think we should do this as a 1-off change.

[mathiasbynens]

We can discuss the tradeoffs of each in the editor call, but right now I don't think we should do this as a 1-off change.

Just to be clear, I agree with this! As I said:

If the editors are open to this change, I’d be happy to work on a patch that renames N as well (perhaps to WithNamedCaptureGroups).

The reason I proposed doing it as two separate patches is because this one would (editorially) affect the RegExp set notation spec proposal whereas the N change wouldn’t. But I’d be happy to update this patch to change both at once if that's preferred.

[michaelficarra]

We talked about this in the editor call. Longer term, we would like to link all of the grammar flags to a place that has descriptions. But for now, changing U to something more descriptive is fine. We just ask that flag names not include Flag in the name (even though we understand that here it is referring to the RegExp flags).

[mathiasbynens]

We talked about this in the editor call. Longer term, we would like to link all of the grammar flags to a place that has descriptions. But for now, changing U to something more descriptive is fine. We just ask that flag names not include Flag in the name (even though we understand that here it is referring to the RegExp flags).

Any suggested new name for UnicodeFlag? Perhaps UnicodeMode or WithUnicode (to match the potential WithNamedCaptureGroups in the future)? It would help to know the exact rationale for avoiding "Flag" (because maybe "Mode" is a no-go for the same reason). For context, I'm trying to think of a name that's unique within the spec document, to improve searchability — so I’d rather not go with just Unicode.

[michaelficarra]

UnicodeMode sounds good to me. Two reasons not to use "flag" in the name:

1. A reader is probably more likely to think the word refers to it being a grammar flag than a RegExp flag, and we certainly aren't going to add Flag to the end of every grammar flag.
2. The grammar flags should be referring to the mode that the grammar is in or the thing that is being controlled, not the cause for entering this mode. For example, the Await flag is called Await because it controls the handling of the await contextual keyword. It is not called Async even though it is introduced by the async keyword.

[mathiasbynens]

I’ve gone ahead and updated the patch to use UnicodeMode. PTAL

• A reader is probably more likely to think the word refers to it being a grammar flag than a RegExp flag, and we certainly aren't going to add Flag to the end of every grammar flag.

That's interesting. I never thought of these as "grammar flags". The spec refers to them as grammar parameters.

[markusicu]

I suggest changing the PR title to say "UnicodeMode"

[mathiasbynens]

Is anything blocking this? I'd be happy to resolve the conflicts if that moves things along.

[michaelficarra]

@mathiasbynens That won't be necessary. We just need @bakkot or @syg to find some time to do a review.

[bakkot]

This LGTM, sorry for the delay. I've also rebased, so it should be good to land.

[ljharb]

@bakkot it looks like the rebase ended up with the wrong commit - d03e65a instead of a rebased b025795.

[bakkot]

Whoops, fixed.