I found a couple locale strings which should work in JS, and they work in Node.JS v14.15.1, but not in Chrome or Firefox. I’m not asking for debugging advice. Where do I raise this issue? Should I go to w3c, or chromium or gecko, or ECMAscript, etc. Is there a GitHub Repo, or do I use a mailing list, etc.
The folks on Chromium’s bugtracker will likely mark it for triage and then ignore it for the next few years (the scars still hurt), but it’s basically here - https://bugs.chromium.org/p/chromium/issues/list
For Firefox, see https://support.mozilla.org/en-US/kb/file-bug-report-or-feature-request-mozilla.
If anyone is interested, the bug is this:
TL;DR
en-US-u-nu-hebranden-US-u-nu-romandon’t have any effect, but-araband-mongdo.The locale string
en-US-u-nu-hebrspecifies that we’re using the Hebrew numbering system (ref). Therefore,new Date().toLocaleDateString("he-IL-u-nu-hebr")should output Hebrew characters. In my browser, it’s “11/24/2020”. In Node.JS, it’s “י״א/כ״ד/ב׳כ׳” (that’s Hebrew).According to MDN,
hebrandromanaren’t valid values wherearabandmongare.Yeah, I saw that. It says “Possible values include”, though. Also MDN says elsewhere that it is valid
Mozilla has a public matrix chat instance at chat.Mozilla.org. For those questions I’d reach out to the localization folks, likely in #l10n, but otherwise folks in #introduction can point you in the right direction
Apparently, the current ECMAscript specification doesn’t support algorithmic numbering systems (GitHub issue). I think the argument is that it would be complicated to implement these. I don’t know why it works in Node.js.
Node.js presumably passes it off to ICU, whereas the browsers do it all in-house.