James M Snell wrote:
>
> Sigh... with special thanks to Robert S. who pointed out to me in a
> offlist note that I completely, totally, and embarassingly forgot about
> that wonderful little thing known as the If-Modified-Since header, I
> gotta say that I'm feeling wonderfully silly at the moment. I'll blame
> it on an acute lack of sleep brought on by the arrival of Snell Kid #3.
> ... in any case, just pretend that I was talking about the
> If-Modified-Since header all along and go from there.
I think I proposed something along this line some months ago, or at least
I thought about it ;-)
I think however I really made this public on the list and someone pointed
me at RFC3229 w/ feed (how would I have discovered it otherwise? ;-) ).
I'm not sure it's good to return something different depending on
If-Modified-Since. If-Modified-Since (or If-None-Match, which would then
be equivalent to RFC3229 w/ feed) is meant as a "give me nothing if hasn't
changed, no need to waste bandwidth with something I already have", not as
a "give me something if it has changed, eventually something different
depending on the date I use" (i.e. if the last change is 3 days ago,
"If-Modified-Since 5 days ago" and "If-Modified-Since a week ago" should
return the same doc).
Exerpt from RFC2616 [1]:
b) If the variant has been modified since the If-Modified-Since
date, the response is exactly the same as for a normal GET.
This brings us back to RFC3229 which explicitly deals with deltas, and
RFC3229 w/ feed (not official though :-( ) which apply them to "records"
(entries in feeds) rather than bytes (as an ediff algorithm would do)
[1] http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html#sec14.25
--
Thomas Broyer
