Relativetimeformat - inconsistency module name for Norwegian

[Ariel-Rodriguez]

Hello,

I noticed that translations for Norwegian are generated with a particular module name for intl-relativeformat.
The module under locale-data is found as nb.js (standard code BCP 47) instead of no.js (standard ISO 3166-1).

I found the translations for intl-relativeformats in the following path:

@formatjs/intl-relativetimeformat/locale-data/nb.js

As meanwhile the translations for intl-pluralrules are in ISO 3166-1

@formatjs/intl-pluralrules/locale-data/no.js

Causes

This inconsistency breaks the pattern and dynamic loader based on locale, hence i am forced to make hardcoded patch

    if (locale === 'no') {
      // the polyfill name for Norwegian is being named with BCP 47 Code
      await  import('@formatjs/intl-relativetimeformat/locale-data/nb');
      return;
    }

   // locale ISO 3166-1
   await import(`@formatjs/intl-relativetimeformat/locale-data/${locale}`)

Why

I would like to know if that module name nb is expected and why differs from other packages.

Does anyone know if my workaround is appropriate or what is the right way to make it work?

Thanks in advance.

[longlho]

nb is available in both:
https://unpkg.com/browse/@formatjs/intl-relativetimeformat@8.1.2/locale-data/nb.js
https://unpkg.com/browse/@formatjs/intl-pluralrules@4.0.11/locale-data/nb.js

[Ariel-Rodriguez]

Hey thanks for the quick reply @longlho

Why is it called nb.js instead of no.js, is there any standard or pattern behind ?

Unfortunately nb does not complies with ISO 3166 (ISO 3166-1 Alpha-2 code).

What standard should i follow to guarantee that all other languages could be programmatically calculated based on country code?

[longlho]

iso3166 are country codes. Languages are from IANA. I think you mean iso639 which mark nb as deprecated since 2000?
I don't think you should programmatically calculate them. They are locale with potentially script & region (eg zh-Hant-TW). Pluralrules only cares about the language itself but others do care about script & region. Also we don't have data for every locale.
What you're talking about is called locale negotiation

[Ariel-Rodriguez]

Alright, locale negotiation is a good suggestion, and solves the problem.

That makes me think that all files under locale-data are being named using BCP 47 tags names maintained by IANA

Besides notes: somehow i still don't understand the inconsistency regarding file naming. I got confused because Norwegian is also named with another name pluralrules/locale-data/no.js which seems not a BCP 47 tag that seems to be a country code tag no. Why is that?

[longlho]

Both are valid language tag: https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry