Re: lynx-dev page counters

[Leonid Pauzner]

7-Dec-98 22:43 David Woolley wrote:
>> > ++ Strictly speaking, it should be made to appear non-cacheable to HTTP 1.0
>> >    but actually be marked must-revalidate, so that HTTP 1.1 software sends
>> >    an If-Modified-Since request, but doesn't require the actual content 
>> > back.
>> and so the HTTP 1.0
>>

> I-M-S is new to HTTP 1.1.  It just happens that it is legal to include some

Berners-Lee, et al           Informational                      [Page 2]

RFC 1945                        HTTP/1.0                        May 1996


       10.9  If-Modified-Since ................................... 42
       10.10 Last-Modified ....................................... 43

> HTTP 1.1 headers in a basically HTTP 1.0 request.  Must revalidate won't
> be recognized by any HTTP 1.0 cache that I know.

The true that I-M-S is not enough for exact validation
because of content negotiations, user agent, etc.;
"Etag" was introduced in HTTP 1.1 for that purposes.
Also the clock inaccurance may be a problem for age calculation;