Manish Goregaokar

I think a really good proposal that gets maximum space wins and is compatible with future map abstractions would be: - We make the VarZeroSliceUnbound abstraction https://github.com/unicode-org/icu4x/issues/5378 - We use...

For 2.0: Audit all stabilizing data structs for: > Finely sliced data markers, defined as a data marker with a large or unbounded number of data marker attributes, should be...

I basically never did this because the ideal way of doing this avoids doing the additional lookup on the vector size, which means different (but faster) lookup code. But we...

https://github.com/unicode-org/icu4x/issues/3128 is not required for, but would help with this

Fixing this changes data representation for anything using MultiFieldsULE. Let's try to fix this for 2.0

If we end up unable to get to this, we punt till next major release, OR create two MultiFieldsULE types where one has the old behavior for compat (and provide...

https://github.com/unicode-org/icu4x/pull/5593

We should try to update to a newer (but pinned) nightly here if possible but I don't think it's super important

r? @eggrobin for the algorithm

Yeah I expect so, this is an extremely aspirational issue. It would involve some refactoring.