Shane F. Carr

A single UnknownMonthCode error variant was objectively an oversight, since we have a use case for it now that we didn't have last year. So, there's already not a superset...

Sorry but what's the "churn"? If it's people migrating to from-fields in 2.1, this isn't measurable at all. If we want to reduce churn, then we should use the nested...

Ok, aside from the consistency and churn arguments, It's completely reasonable for code to want to have a code path that works in terms of RangeError. There are multiple APIs...

Let's say for the sake of argument that the consistency and churn arguments cancel out, since there are potentially reasonable positions on both sides. Looking then at technical merit: Advantages...

Good point, I didn't consider that. This type has a lot of functions, like `load_*` and `include_*`, and it seems nice to maybe group these ones as `get_*`. But, it's...

Note that there are two places where floating point numbers are rounded: first in the conversion of the floating point number to the shortest decimal string, and second when performing...

Working group discussion: - @sffc We've looked at the problem space. What is the overall direction we want to go? - @echeran What does ICU4C do if you call the...

I'm okay adding a `Month` type that is similar to the internal type `ValidMonthCode` and using it in `try_from_codes` and potentially adding it to `MonthInfo`.

#7147 landed. Things I'd like us to follow up on: - Constructors on the Month type - Range of months to allow in the Month type - Documentation around the...

Sorry if I wasn't clear; this would be a new type in the zerotrie crate, named something like `ZeroReversableTrie`.