Henri Sivonen

I don't recall seeing demand for these encodings for Web compat purposes, so I think we shouldn't be adding these. I don't recall seeing evidence of `x-mac-cyrillic` needing to be...

I started making borrowed versions of the normalizer structs. I pushed my non-compiling WIP to https://github.com/hsivonen/icu4x/tree/borrowednormalizer in case anyone wants to take a look while I'm away. I got stuck...

> I think it shouldn't block 2.0 unless we can come up with a way to do this by tweaking the DataPayload architecture somehow. That seems like a much bigger...

Normalizer PR in #5413 along the lines of the previous comment here.

This is done for normalizer and collator. This is not done for segmenter (#5514). Not sure about other components.

> > AFAICT, the only way for `CollatorBorrowed::try_new` to fail is to override baked data with badly-generated baked data. > > No, you can give it bad options. What bad...

`unicode-ident` cites https://github.com/rust-lang/rust/pull/33098/files for a data structure that's similar to `CodePointTrie` with single-bit values. In terms of goal formulation, if we want `icu_properties` to be the single source of Unicode...

> What is the "primary provider-backed payload"? It's not well defined in all cases. For the collator, it would be the payload for `CollationMetadataV1Marker`. > The whole concept of the...

> > Either way, which key should the glue code query for if the primary payload for a given ICU4X object isn't well defined in all cases? > > I...

> Would it be incorrect to always return the requested locale as the resolved locale? That question can be understood in at least three senses: 1) what the caller wants...