Re: Issue List: CACHEDATE

Larry Masinter wrote:

> > I can contribute an application which causes I-M-S dates differing from
> > the original modification time. Assume that documents are created on the
> > fly by a server/CGI script using multiple other documents or some
> > external data. The overall modification time is computed as the latest
> > modification time of all parts.
> 
> > If the browser uses this modification time for an I-M-S request and if
> > one of the document's parts is fetched using a proxy mechanism the
> > original modification time of this part is no longer available. Thus,
> > the I-M-S time used in requesting this part may be later than the
> > modification time for this part and, refering to the example above, the
> > server should return 304, because this part has not changed.
> 
> I'm confused by your example. If the parts are not modified, then they
> have the same date as they had the first time. If they ARE modified,
> then they have a different date. If you can't tell whether the parts
> are modified, then you can't tell the modification time.

Without an I-M-S header a requested document is composed out of parts. The
modification time for the composed document is computed as the latest
modification time of all parts.

With an I-M-S header the parts are checked whether they changed since the
modification time of the composed document. If one of the parts changed,
the document is composed as usual. If not, a 304 response is generated.

If one of the parts is fetched from another server using HTTP again,
then we would like to use again an I-M-S header to check the part's
modification time. We only could use the same I-M-S header as for the
composed document because the original modification time of the part is no
longer available. So, the I-M-S time is the modification time of the composed
document which is equal or *later* to that of the part originally retrieved.

In this case we would like to know whether the document changed since exactly
*this* time. At *this time* it was apparently not modified as we used it to
compose the document.


--Franz