Henri Sivonen

> Why can't we just expose sensible utility functions like `startsWith` and `indexOf` directly on the Intl.Collator that just does what the builtin JavaScript string functions do. It's just UTF-16...

> Can `Intl.Collator` just expose methods for iterating through collation elements? It seems like a very bad idea to expose this kind of implementation detail in a way that would...

> Collation elements are the technical equivalent of the everyday concept of letters of an alphabet. But they aren't. There might be more or fewer collation elements than letters (for...

CC @mohmoto

While at it, it would probably be a good idea to clarify what the language in the README about both MIT and Apache means. I expect it not to mean...

I think the last option is the ideal one. A quick look at clang's arm_neon.h suggests that 1. The multiple vectors wrapped by a returned C `struct` are always mutually...

A tuple is nicer for Rust destructuring. (Also, since the ARM stuff is already strongly-typed, `uint8x16_t` could be `u8x16` from portable SIMD.)

* `decode` _changes the encoding that's used_ according to the BOM and removes the BOM. * `decode_with_bom_removal` removes the (leading) BOM if it matches the BOM for the encoding but...

> Furthermore I don't see any documentation indicating that `decode()` removes the BOM, could you point out where it is documented? Good point. It says "BOM sniffing" without saying explicitly...

The BOM isn't part of the XML information set, so an XML use case hasn't arisen before. Does you exact reproduction mode also retain whitespace on either side of the...