Re: [xri] comments on Expires

[Will Norris <will@willnorris.com>]

On Jan 13, 2010, at 7:52 AM, Scott Cantor wrote:

> Nika Jones wrote on 2010-01-13:
>> ...  A resolver MUST NOT use an XRD after the time stated here ... If the
>> HTTP transport caching semantics specify an expiry time earlier than the
>> time expressed in this [element], then a resolver MUST NOT use this XRD
>> after the expiry time declared in the HTTP headers per section 13.2 of
>> [RFC2616] ...
>> 
>> I think that says it all.
> 
> Except that cache control != validity, and XRD Expires is about validity.
> 
>> I suggest that anytime after the <Expires /> date has past, that the XRD
>> MUST NOT be used, regardless of HTTP caching. Because as you mentioned the
>> same document should be served up (hence cache.)
> 
> It's nothing to do with HTTP, it's just that if it's Expired, it isn't valid
> any more and you're basically using it at your own risk. Normally specs stop
> short of saying MUST NOT use. Usually it's phrased more as the data in the
> XRD has no presumed validity past that time.
> 
> The HTTP caching layer, if in fact HTTP is used, just controls the
> conditions under which a new copy might or might not be obtained, and at the
> end of the process, you either have a valid XRD or an expired one.
> 
> -- Scott

what do you thing about Joe's point about the confusion of the word "valid" in the context of XML validity?  Would it be safe to say that if an XRD document has expired, then consumers SHOULD attempt to obtain a new copy?  Or do think that is still overstepping the scope of the specification?  I tend to agree with Joe and others that *some* direction should be given, whatever the strength of the language is.  If we are going to talk about the document being "invalid", then we should give some indication of what the means, and what action should be taken.  ??

-will