Re: XMLHttpRequest conformance comments

Ian Hickson schrieb:
> On Wed, 2 Aug 2006, Julian Reschke wrote:
>> If compatibility to existing code (which doesn't check for null) is the 
>> driver here, then please consider adding a new method such as 
>> "isHeaderPresent(headername)".
> 
> Purely out of interest, could you give the use case?

Only for a related problem: early versions of the WebDAV REDIRECTREF 
extension defined a header like that 
(<http://greenbytes.de/tech/webdav/draft-ietf-webdav-redirectref-protocol-06.html#rfc.section.11.2>). 
It was subsequently changed because it couldn't be used with Microsoft's 
XmlHttpRequest object. I wouldn't be surprised if similar headers have 
been defined by others (but possibly not published through the IETF 
process).

Speaking of which, it might be good if 
<http://www.w3.org/TR/XMLHttpRequest/#dfn-setrequestheader> would 
clarify that setting a request header to "" is legal, and that the 
header really MUST be set to that value.

Finally, looking at the current text over there:

"# Nothing MUST be done if the header or value  arguments contain any 
U+000A LINE FEED or U+000D CARRIAGE RETURN characters, or if the header 
argument contains any U+0020 SPACE or U+003A COLON charecters.
# Nothing MUST be done if the header argument matches Accept-Charset, 
Accept-Encoding, Content-Length, Expect, Date, Host, Keep-Alive, 
Referer, TE, Trailer, Transfer-Encoding or Upgrade case-insensitively."

This is extremely misleading. "Nothing MUST be done..." essentially 
means that everything is optional here. I think what it intends to say 
is something like... "The header MUST NOT be changed if....".


Best regards, Julian

Received on Wednesday, 2 August 2006 08:28:46 UTC