Shane F. Carr

Just some thoughts on the weekday thing; this diverges a tiny bit from UTS 35, but: ``` # Year/month/day {$input :date fields=YMD} # Year/month/day with weekday {$input :date fields=YMD weekday=true}...

What I mean is a _developer configurable_ day period length. If a locale doesn't have am/pm, then that locale can still opt to use the flexible day periods.

I'm not 100% happy with what got landed as draft and want to continue to explore this space, so I think we should keep this issue open.

I'm hitting this error in ZonedDateTimeFormatter FFI, #6233 It gets triggered because I first load the names into the DateTimeNames with the helper functions and then call `try_into_formatter` which calls...

I worked around the case hit in #6233 by relaxing the check for `is_none()`. If all three loads return `None`, then we can assume that all of the payloads were...

I'm not sure about the best way to resolve this issue and whether it impacts 2.0 stability.

@Manishearth suggested a solution (I thought in a comment but maybe not): we can save the checksum in the Names but not in the Formatter, so as not to increase...

The philosophy we've been using in datetime input, which has been working well, is that it is the input type's responsibility to transform itself into the inputs the formatter needs....

> > A third-party type that doesn't have metazone information baked into it could use MetazoneCalculator inside of its impl GetInput. > > But it can't, because in the `impl...

> We could then remove the metazone and zone_variant fields from icu::timezone::CustomTimeZone as well. They are already not populated when parsing an IXDTF string, which I consider the main use...