@zbraniecki, any idea?

I've done a tad of research, and found http://cldr.unicode.org/translation/capitalization and http://cldr.unicode.org/development/development-process/design-proposals/grammar-and-capitalization-for-date-time-elements. Which implies that anything on the CLDR/ICU front would return lower-case strings for Hungarian, even in standalone.

Now, I don't know if contextTransforms are exposed in any way. And in Fluent, I think it's hard to do so out of the box.

I wonder if the least horrible solution would be to expose a CAPITALIZE function, and have languages that need it wrap the DATETIME call like so: `{CAPITALIZE(DATETIME($isoString, weekday: "long"))}

So, in ECMA402 we're introducing Intl.DisplayNames which is an API better suited for cases like retrieving a weekday.

With that in mind, we're talking about adding displayContext to it - tc39/ecma402#355

This will take some time to standardize and in the meantime, we do not provide a good intl coverage for capitalization rules.
You can write your own, as Pike suggested, but it may require a bit more trickery, since by default we just return data from CLDR for context standalone which may or may not be capitalized, so you need to adapt for both:

{ CAPITALIZATION(DATETIME($isoString, weekday: "long"), "lower") }
{ CAPITALIZATION(DATETIME($isoString, weekday: "long"), "title") }

etc.