Shane F. Carr

Initial thoughts: 1. OK with Proposal 1 2. This is a somewhat radical change to the mental model, but I can buy into it because I've never been a huge...

We could have both inlined const fields and a helper `DataMarkerInfo` ```rust pub trait DataMarker: DynamicDataMarker { const NAME: DataMarkerName; const IS_SINGLETON: bool; const FALLBACK_CONFIG: LocaleFallbackConfig; } #[non_exhaustive] pub struct...

- @Manishearth - Overall I like merging the names. But the "Marker" name is overloaded. - @sffc - Marker is a common name that's used for the type-system level objects...

Notes from the scratch pad: ```rust pub fn try_new_unstable( provider: &D, locale: &DataLocale, options: CollatorOptions ) -> Result where D: DataProvider + DataProvider + DataProvider + DataProvider + DataProvider +...

Discussion: - @Manishearth Users typically don't use the string key, but it has its use cases and might show up in errors from time to time. I think the status...

I am assigned this issue for 2.0 to do an audit of datetime markers. [^1] The audit, whether in icu_datetime or elsewhere, brings back the data marker naming discussion where...

> I'm in favor of this. I don't necessarily think we need to _require_ qualification for cross crate stuff. Currently because of `provider` it will lead to long `icu_decimal::provider::SymbolsV1` imports...

Thought: we could make the infixed marker names be `MonthNamesV1GregorianMarker`. Then the path could always be derived directly from the marker name and vice-versa, without having to know the data...

> Yeah this makes a lot of sense to me! I think we prefixed it because we always did, but I actually don't see a strong reason to prefix. Just...

> > Just highlighting that my last reply above was to put the infix between the "V1" and the "Marker"? > > Oh, I see. Hmmmmm. I'd rather have it...