Re: Referer URI MUST NOT include a fragment

2009/3/1 Mark Nottingham <mnot@mnot.net>:
> Well, not *that* portable. I don't think HTTP should be talking about the
> syntactic details of specific media types. Delegating this to media type
> definitions would be the most sensible long-term thing to do, although some
> sort of advice would be needed in the meantime...

I agree; but at the moment the spec says Referer MUST NOT include a
fragment identifier, and the case that I am making is, in essence,
this should be relaxed to SHOULD or MAY. I would expect an implementor
to shoot this proposal down if it means contradicting HTTPbis as it
currently stands.

I see no reason why this proposal could not preserve the delegation of
the semantics of source fragment identifiers to media types. However,
it might be pragmatic for HTTPbis to offer non-normative guidance or
advise on HTML, which I feel would cover a large majority of use
cases. I think this could be done in a single paragraph, paraphrasing
this thread. I would also be willing to draft a seperate patch for
HTML5 if you think that's the way to go, but it would obviously help
if one could point to consensus on HTTPbis.

Received on Sunday, 1 March 2009 01:12:47 UTC