A. Pagaltzis wrote:
No. RFC4287 does not merely recommend it, it RECOMMENDS it.
I don’t know about you, but I consider a SHOULD to be pretty
strong language.
I consider it as strong as its definition which clearly says: "there may
exist valid reasons in particular circumstances to ignore a particular item,
but the full implications must be understood and carefully weighed before
choosing a different course."
I don't take SHOULD recommendations lightly, but I think I've shown some
pretty compelling reasons why a producer (if they really absolutely have to
use "application/xhtml+xml"), would be wiser to use an xhtml document
fragment than a complete xhml document. Of course the wisest move would be
to not use any form of xml at all, but that's not what was being discussed.
Aggregators which process @type='application/xhtml+xml' as if it
was @type='xhtml' are in error. Period.
This argument I don't understand. The spec, while recommending (super
strongly recommending, if you will) that the content should be a full xhtml
document, clearly also allows that "there may exist valid reasons in
particular circumstances [for a producer] to ignore [that recommendation]".
How can an aggregator that makes allowances for such an eventuallity be in
error?
Or are you suggesting that RFC4287 is making some other requirement of Atom
aggregators that I've missed? According to Rob, the spec doesn't actually
require that Atom aggregators do anything with MIME types. I'm not sure I
agree with him completely, but section 4.1.3.3 only tells you what to expect
to find in the content, not what you are required to do with it.
But *the spec* leaves not a shadow of a doubt about what it means
when the type of content is `application/xhtml+xml`.
Thus, content producers who still follow the 0.3 custom are
clearly in error too. Period.
This argument I understand, but I still disagree. I'm willing to accept that
this behaviour is highly discouraged, but it's not a REQUIREMENT and it's
not an error. Try persauding anyone on the feedvalidator mailing list that
SHOULD recommendations should all be flagged as errors and see how far you
get. I'm pretty sure I'm not the only one that will disagree with you.
Regards
James
