Should Charset mention UTF-8

[martinthomson]

Accept-Charset is deprecated due to the ubiquity of UTF-8, but the Charset section makes no mention of UTF-8. This might be worth a mention.

[annevk]

It's only used for Accept-Charset though and that section already mentions UTF-8. As does the Media Type section. Since Accept-Charset is deprecated, perhaps the charset production should simply be removed and Accept-Charset takes token directly?

[royfielding]

Accept-Charset is deprecated primarily for privacy. It also happens to be less useful now because of utf-8. In contrast, the charset parameter is still used most of the time, mainly to reduce sniffing. We can add something about utf-8, but promoting it too hard will cause other (social) problems.

[reschke]

@annevk's observation is correct. But then, charset is used in Content-Type as well, just not referenced directly as it is used over there as a regular parameter.

[annevk]

It's not clear to me at least that the charset production section also applies to Content-Type.

[reschke]

Syntactically, it does not (because that relies on a more generic syntax for arbitrary parameters).

But I agree, this needs cleanup.

[reschke]

Proposal is to remove the ABNF production and to clarify the text (so that it applies to Content-Type as well).