14:42
<eemeli>

I vaguely recall some recent-ish conversation somewhere about defining or codifying the behaviour of "best fit" locale matching, but now I can't find any record of it. Am I misremembering things?

Context here is that I happened to notice that no browser seems to fall back from hak to zh, even though the latter is a macrolanguage for the former, and CLDR encodes that relationship as a languageMatch, and I'd like to figure out what the right place might be for fixing that.

17:20
<Richard Gibson>
not that I recall, but I suspect we could defer to a lot of https://unicode.org/reports/tr35/#LanguageMatching while still allowing implementation-defined weights/thresholds/etc
22:27
<sffc>

I vaguely recall some recent-ish conversation somewhere about defining or codifying the behaviour of "best fit" locale matching, but now I can't find any record of it. Am I misremembering things?

Context here is that I happened to notice that no browser seems to fall back from hak to zh, even though the latter is a macrolanguage for the former, and CLDR encodes that relationship as a languageMatch, and I'd like to figure out what the right place might be for fixing that.

We recently discussed https://github.com/tc39/ecma402/issues/947 which is somewhat related, but I don't recall discussing codification of BestFit