21:02
<sffc>
CLDR performs some fallbacks that aren't allowed by naive subtag stripping in "best fit", like en-GB to en-001 or hi-Latn to en-IN
21:03
<sffc>
However, I don't think this behavior is currently observable because ICU, and therefore browsers, treat anything with a nonstandard fallback to be in the availableLocales list
21:04
<sffc>
The spec reads as if it is based on the UTS 35 skeleton resolution algorithm, which is the same algorithm used in ICU
21:05
<sffc>
For both this and localeMatcher: "lookup", it's not clear to me what the motivation was to write what are essentially reference algorithms in the spec when most everyone will be using the alternative implementation-defined routines anyway
21:06
<sffc>
and since these are data-driven algorithms. People can't rely on the behavior of either localeMatcher: "lookup" or BasicFormatMatcher because they are subject to change when data changes.