Rich, On 19 May 2013, at 06:30, Richard Alimi wrote: > On Fri, May 17, 2013 at 8:13 AM, Ben Niven-Jenkins <[email protected]> > wrote: > > That's basically what I was suggesting except rather than being specific > about what the server must use, leave it to the client to determine based on > how the client calculates freshness (as the client is likely in the best > position to know what rate it might want to poll at as it know how up-to-date > with any changes it might want to be). > > RA: I'm fine with adding a statement such as the following: > > RA: An ALTO Client determines the frequency at which ALTO information is > refreshed, and MAY be determined based on freshness information made > available via HTTP. > > RA: How does that sound?
Honestly I don't think it adds anything as this is what a client is likely to do in the absence of any text in the spec. > If we're going to pick one mechnaism I;d prefer we pick Cache-Control: > max-age myself rather than Date/Expires (as max-age takes priority over > Expires according to RFC2616). > > RA: I don't think that ALTO is the place to recommend one or the other. If > this is desired, it seems like this should be something in httpbis (if it > isn't already), or a different draft independent of ALTO. I agree we shouldn't try to pick one versus the other (hence my original suggestion was to base it on the freshness of the response rather than a particular freshness mechanism) however I do think it would be useful for the ALTO spec to say something. Atom didn't specify anything for this case either and it caused a lot of confusion for deployers/implementors looking at the questions on stackoverflow/etc. If we want to avoid RFC2116 type words (which would apply to this case IMO as it is a form of interop issue) then maybe something like: An ALTO Client determines the frequency at which ALTO information is refreshed and it is suggested that ALTO clients do not refresh ALTO responses before they become stale according to RFC2616. Ben _______________________________________________ alto mailing list [email protected] https://www.ietf.org/mailman/listinfo/alto
