Make PositionEncodingKind a type

[Michael Peyton Jones (michaelpj)]

As discussed here:
f9c85d5#r70968669

I made a change that reflected the existing state, but looking at it, it still reads somewhat ambiguously to me.
We are talking about an overall "position encoding", but in fact it only has quite a specific meaning: what is counted by character offsets.

If we were open to a slightly bigger change, we could consider something like:

// with a proper type obviously
type CharacterOffsetUnit = 'utf-8-codeunits'|'utf-16-codeunits'|'unicode-codepoints';
interface PositionCapabilities {
  characterOffsetUnits: CharacterOffsetUnit;
}

[Deleted user (ghost)]

All CLA requirements met.

[Felicián Németh (nemethf)]

Maybe I'm overthinking this, but explaining the original a𐐀b example of the current specification for each encoding would make this even clearer. Or not. Thank you.

[Michael Peyton Jones (michaelpj)]

Maybe I'm overthinking this, but explaining the original a𐐀b example of the current specification for each encoding would make this even clearer. Or not. Thank you.

I think so long as we're referring to clearly understood ways of dealing with text, we can assume that programmers will go elsewhere to understand what e.g. "indexing in UTF-8 code units" means.

[Dirk Bäumer (dbaeumer)]

I think so long as we're referring to clearly understood ways of dealing with text, we can assume that programmers will go elsewhere to understand what e.g. "indexing in UTF-8 code units" means.

Michael Peyton Jones (@michaelpj) are you willing to provide this as a PR as well?

[Michael Peyton Jones (michaelpj)]

I meant that I think the current text is clear enough, since it uses standard terminology. Perhaps Felicián Németh (@nemethf) disagrees?

[Dirk Bäumer (dbaeumer)]

Michael Peyton Jones (@michaelpj) sorry misunderstood.

[Felicián Németh (nemethf)]

Perhaps Felicián Németh (@nemethf) disagrees?

I don't.