Shane F. Carr

Okay, as demonstrated in #4929, increasing the size of the offset above 32 bits is unavoidable. Given the increased bits, I've now implemented a solution that pins the winter solstice...

I already pulled out some of the changes, but if desired, I can split this PR up further.

Branch with individual commits archived at https://github.com/sffc/omnicu/tree/archive/negative-tests

In any of these modes that involves boxing the data struct for postcard users, it would be nice if they could work with a cache of deserialized postcard objects (the...

Discuss with: - @Manishearth - @sffc - @robertbastian

I don't see a way to get to 16 for a cowable slice. In the non-null case, both the borrowed and the owned variants need two full usizes. The only...

_Maybe_, but then we technically need fallible conversions from `Box` and `&str` into `HalfLengthCowStr`. Or maybe that's enough of a degenerate case that we can debug-assert and GIGO.

This is fixed by https://github.com/unicode-org/icu4x/pull/4904

This is a good first issue. Adding it should be similar to adding `Hangul_Syllable_Type` which I did last week: https://github.com/unicode-org/icu4x/pull/4885

@Manishearth Anything left on this issue?