RFC 7231, "Hypertext Transfer Protocol (HTTP/1.1): Semantics and Content", June 2014Source of RFC: httpbis (app)
Errata ID: 4031
Reported By: Anne van Kesteren
Date Reported: 2014-06-30
Rejected by: Barry Leiba
Date Rejected: 2014-07-01
Section 126.96.36.199 says:
media-type = type "/" subtype *( OWS ";" OWS parameter )
It should say:
media-type = type "/" subtype *( OWS ";" OWS [parameter] )
See the thread at http://lists.w3.org/Archives/Public/ietf-http-wg/2011JulSep/0027.html
Implementations are much more relaxed when it comes to parsing MIME types.
The above is probably still too strict. E.g. requiring that a parameter contains "=" is something I doubt is actually the case in practice.
The ABNF is there to specify what the expected productions are, and is correct as it stands: we do not *want* things such as these to be produced:
Content-Type: text/plain; charset=iso8859-1;
Content-Type: text/plain;;;;;;;;;;;; ;;; ;;; ;;;
That said, this report addresses a real problem: lack of direction to parsers on how to be appropriately lenient. Because the fact is that general interoperability of this stuff in the wild would be improved if parsers accepted at least the first two of the examples above, which are illegal by the grammar, but which do get generated by less-than-perfect implementations.
I'm marking this report as "Rejected" because the problem it means to address is much broader than this one case, and can't be fixed with an errata report. But it's important that we take up work on a document that does properly address this issue.