Html: should it be called "case-sensitive"?

[aphillips]

https://html.spec.whatwg.org/multipage/infrastructure.html#case-sensitivity-and-string-comparison

When discussing issue #5066 members of the I18N WG expressed concern that the use of the term "case-sensitive" was misleading, since what is actually happening is codepoint-by-codepoint comparison. Should "case-sensitive" be called something else, such as "identical" or "codepoint"?

Note: I'm aware that the term "case sensitive" is already used and linked widely, not only in HTML but also in many attendant specs that refer to the definition provided by HTML.

[annevk]

I'd be okay with minting "equals" for this (or perhaps "is", since I suspect that's generally what we use these days), in https://infra.spec.whatwg.org/#strings. But someone would have to do the work of identifying all specifications and getting them all changed.

[Yay295]

"visually equivalent"? Though I suppose that would have other issues.

[aphillips]

@Yay295 There are plenty of visually equivalent strings that should not be treated as equivalent. Cf. here

[Celedence]

"case dependent" Might be a good way to go too?

[aphillips]

I reviewed the current use of case-sensitive in HTML. There are 39 occurrences of it (not counting the 2 in the definition). In a number of cases, the phrase is used to emphasize that value-matching expects string equality, especially with regard to case. And this is, obviously, only in the HTML spec. Before making PRs, I thought I ought to check on the right course of action. What I had in mind was:

1. In Infra, define is as code point-by-code point comparison. Include a synonym identical to for cases where the spec needs to emphasize the equality.
   a. Include a note emphasizing that this is case and code point sequence (normalization) sensitive.
2. In HTML replace case-sensitive with a reference to is.
   a. Include a note explaining that this was formerly called case-sensitive and retaining the anchor/definition for other, yet-to-be-modified specs.
3. Replace the 39 occurrences with suitable edits to is/identical to (or whatever we decide to spell them as)

Does this sound like the right approach?

[domenic]

Generally sounds good to me.

In Infra, define is as code point-by-code point comparison. Include a synonym identical to for cases where the spec needs to emphasize the equality.

Probably you'd also want to include a note that by default, specs using Infra compare strings in this manner. I don't want people to think that if a spec compares strings without linking to "is" or "identical to", then it's undefined. (This would basically be a port of HTML's "Except where otherwise stated, string comparisons must be performed in a case-sensitive manner.")

a. Include a note explaining that this was formerly called case-sensitive and retaining the anchor/definition for other, yet-to-be-modified specs.

Since HTML doesn't have automatic cross-linking anyway, we'd have to add a bullet point for the relevant "is" to the Infra section of https://html.spec.whatwg.org/#dependencies. At that point maybe you should just add the appropriate anchor there (probably as an empty <span id="case-sensitive"></span>). That seems cleaner than having a leftover <dfn> in https://html.spec.whatwg.org/#case-sensitivity-and-string-comparison.

At that point https://html.spec.whatwg.org/#case-sensitivity-and-string-comparison will be almost deleted; the remainder would just be the uses of "prefix match" which are in AppCache and which @annevk preferred to leave alone until we remove AppCache.