review: various timezone/language/format results

[Thorin-Oakenpants]

https://arkenfox.github.io/TZP/tests/timezonenames.html - terrible perf, wish I could do a combined result

So anyway

If you're not hiding your language, it seems as if two languages can still affect my TZ RFP notation .. and now I just realize I was using resolvedOptions 🤦

@abrahamjuliot I'm tired, do you want to whip up a test to return the resolvedOptions timezone for all locales - thanks

[Thorin-Oakenpants]

I am referring to this

[abrahamjuliot]

I'll take a closer look at this. I plan on focusing more on Intl methods. I'm way behind on that.

[Thorin-Oakenpants]

fyi: FF85+ 1635561 this changed Moz.Intl to use RegionalPrefLocales rather than AppLocale, and this is where the system locale kicked in

The only way to protect this is for en-US only, which is a bit of a hole IMO, that RFP could tighten. TB I think is OK when they hit ESR91

[Thorin-Oakenpants]

update:

• I went here and grabbed the raw 4mb+ langtags.json file: date 2022-12-15
• I ran a script to return `{ name:, [tag + tags] } - this gave me 9413 locales
• I then a script on that result to only return the first tag per name (otherwise we're duplicating with aliases or whatever you call them)
• I then filtered the tags (locale codes) through supportedLocalesOf and got down to 1000 and change
• I also removed a few that simply map to other codes, and added those other codes if they were missing
• I then merged it with my base list (which includes all ISO_639-1 alpha-2 codes: not all are supported in FF), de-duped it all

So I now have

• 284: a base core list comprising only 2 and 3 letter codes (well, it also has 6 zh-* codes)
  • all PoCs filter for supported locales, so only having 226 in firefox is fine, it allows us to notice new core additions and prompt us to check for new the monster list
• 839: an expanded list comprising 2-* and 3-* letter codes
• 1123: total - of which FF supports 1065 (775 in collator)
  • I use this list to add extra locales to each PoC to illustrate the max entropy

I can flip the base list to the full 1123 globally, and rerun all the PoCs to see if the number of buckets change

So I think that's about it for covering the field, no need to dig further. I mean there's almost 9.5k listed by CLDR, and I don't think that's all of them that are possible

[Thorin-Oakenpants]

new summary: so this is out of 1065 supported locales

FF125+ (ICU 73 still?)

updated: Aug 2024 FF131

          collation   84
          dayperiod  210
         listformat  140
         compact nf  158
         toparts nf   44
    resolvedoptions   58 (calendar, caseFirst, day, numberingSystem etc)

        currency nf  407 (95 currencies) <- max entropy
        currency nf  275 ( 7 tests, 4 currencies) <-- TZP uses this

     datetimeformat  301 (date-time components: excludes timezones or actual "full" dates)
   date-&-timestyle  321 (actually 321-323 depending on your timezone)

        notation nf   74
            sign nf   38
            unit nf  191

        pluralrules   46
 relativetimeformat  238

       timezonename  311 (40 timezones) <- max entropy
       timezonename  254 ( 5 timezones) <-- TZP uses this
            ------------------
           combined