- From: Julian Reschke <julian.reschke@gmx.de>
- Date: Wed, 27 Aug 2008 16:26:23 +0200
- To: HTTP Working Group <ietf-http-wg@w3.org>
Hi, as discussed last week I have been working on a test suite checking the UA support of RFC2231 encoding -- <http://greenbytes.de/tech/tc2231/>. One thing I wasn't sure about in <http://greenbytes.de/tech/webdav/draft-reschke-rfc2231-in-http-latest.html> was where exactly it makes sense to allow whitespace around the parameter value assignment -- which, as far as I can tell, is not really clear from RFC 2231. Consider these test cases: (1) Content-Disposition: attachment; filename *=UTF-8''foo-%c3%a4.html (2) Content-Disposition: attachment; filename*= UTF-8''foo-%c3%a4.html (3) Content-Disposition: attachment; filename* =UTF-8''foo-%c3%a4.html (<http://greenbytes.de/tech/tc2231/#attwithfn2231ws1>, <http://greenbytes.de/tech/tc2231/#attwithfn2231ws2>, http://greenbytes.de/tech/tc2231/#attwithfn2231ws3>) Both Opera and FF understand variants (2) and (3), but FF chokes on (1). For my RFC2231-in-http proposal I see several choices: a) conservative -- do not allow whitespace in or around the assignment; this is interoperable, and I really don't see why one would *want* to add whitespace, b) pragmatic -- allow exactly those cases where interoperability is there, c) ambitious -- allow all cases, and hope that implementations will be fixed and become interoperable. For now I'm leaning to a) in the spirit of keeping things as simple as possible. BR, Julian
Received on Wednesday, 27 August 2008 14:27:08 UTC