Shane F. Carr

> > struct DTFPreferenceBag + provider + options bag (what is being proposed) > > This is not what I propose here. I propose just `DTFPreferenceBag + provider`. > >...

Note: my envisioned data model for DataLocale is a Language Identifier + map from Unicode Extension Key to Unicode Extension Value. (This is exactly what ResourceOptions is currently.)

> > Where do non-preference options fit in, like date style, components bag, etc? (This is what I meant by "options bag") > > They are part of `DTFPreferencesBag`. `Locale`...

Thanks for explaining your mental model more. I think it is a valid model, and a perspective to consider, but I'm not yet convinced that it is superior to the...

> Architecture of an API should be designed along the axis of fallible/infallible and lossy/lossless. Locale as is is fallible and lossy compared to Preferences. I disagree with this statement...

Discussion: - @echeran - The proposal creates a whole lot of static typing. Not sure it's worth it. Not a fan of `loc.into()` because it's very magic; `From` is more...

Thanks for the contribution! Can you provide figures on how "cargo bench" differs before and after this PR? Boxing the PluralRuleList seems like a good idea. It's one heap allocation,...

Depends on #140 Depends on #107

Now that #1240 has landed, should we close this PR and the corresponding ticket as obsolete?

> Personally I'm in favor of this dependency; it's a build-time only dep and it doesn't add deps other than itself. @sffc ? It's a fine dependency to add from...