Re: [IndexedDB] Add locale-aware sorting (#38)

Locales are not necessarily going to be consistent across browsers. The locales available to a given browser depends on the platform it is running on and/or implementation choices by the browser vendor. Not all browsers use ICU internally, for example. And different locales may or may not be provisioned on different host environments.

CLDR and the Unicode Collation Algorithm (UCA) do provide a basis for building and tailoring collations. These do evolve over time as the data is updated, new characters are added to Unicode, and so forth. Classical database vendors (such as Oracle for example) often go to great lengths to preserve existing collations in a stable manner. This is a good thing for their customers, but may not be suitable for the Web platform. Note that there are ways to tailoring ICU's implementation, at least, to be consistent (by providing the collation rules).

Also, UCA and CLDR describe additional settings (which can appear in a locale identifier courtesy of the 'u' extension of BCP 47 (cf. RFC 6067 and UTR#35)) beyond those in what you might be thinking of as "the locale" and these are also provided in the JavaScript internationalization extension. The JS folks felt that collation stability or even consistency was less important than being as close to end-user expectations at possible.

---
Reply to this email directly or view it on GitHub:
https://github.com/w3c/IndexedDB/issues/38#issuecomment-142068056