Editorial: Rename Unicode encoding/decoding operations

[bakkot]

Fixes #1863 following @domenic's suggestion there. I've also renamed UTF16DecodeSurrogatePair to UTF16SurrogatePairToCodePoint and UTF16DecodeString to UTF16CodeUnitsToCodePoints. I'm open to other names; these seemed the clearest.

After just renaming, some of the wording became a little strange: "the UTF16CodePointToCodeUnits of a Unicode code point" reads more awkwardly to me than "the UTF16Encoding of a Unicode code point" did. So I rephrased those usages.

cc @jmdyck

EDIT: names changed.

[jmdyck]

Hm. The names UTF16CodePoint[s]ToCodeUnits suggest that there's such a thing as a "UTF16 code point", which there isn't. Maybe CodePoint[s]ToUTF16CodeUnits.

[bakkot]

Ah, I'd intended it to read as "the UTF-16 version of code point to code unit translation", but I can see the ambiguity there.

[jmdyck]

(It's tempting to moreover suggest replacing UTF16CodeUnits with just String:

• CodePointToString
• CodePointsToString
• StringToCodePoints

but I haven't actually determined if that's a good suggestion. It does seem more readable.)

[mathiasbynens]

Came here to say what @jmdyck already said: UTF16CodePointToCodeUnits → CodePoint[s]ToUTF16CodeUnits, at a minimum. That said, I like the suggestion to just use String even better.

[bakkot]

I am concerned about saying String instead of CodeUnits because the spec currently makes a distinction between those: StringValue of StringLiteral, for example, says

Return the String value whose code units are the SV of this StringLiteral.

---

Actually, I guess that means UTF16CodePointsToCodeUnits and UTF16CodeUnitsToCodePoints are bad names, because those do return and take a String value respectively.

So, thoughts on

• UTF16Encoding -> CodePointToUTF16CodeUnits
• UTF16Encode -> CodePointsToString
• UTF16DecodeSurrogatePair -> UTF16SurrogatePairToCodePoint
• UTF16DecodeString -> StringToCodePoints

?

[jmdyck]

I am concerned about saying String instead of CodeUnits because the spec currently makes a distinction between those: StringValue of StringLiteral, for example, says

Return the String value whose code units are the SV of this StringLiteral.

Yeah, but I think that's a pointless distinction. See Issue #828.

(I started a PR to dissolve that distinction, but something stopped me. I'll have another look.)

[bakkot]

Force-pushed commits using the names above.

@jmdyck

Yeah, but I think that's a pointless distinction.

Arguably, but if we do away with it we should do so holistically. We can rename CodePointToUTF16CodeUnits to CodePointToString at that point.