Robert Dahlem wrote:
> What I don't understand:
>
> - Why the heck isn't Netscape satisfied with "not modified" and shuts up
> but instead requests a byte range of the same document in the same
> second?
Because the first is a validation to see if the partial response (bytes 0-984)
Netscape has is still up to date.. a crude method, but HTTP/1.0 does not
leave many choices.
The second request sent by Squid is a If-Modified-Since request because the
object already exists in the Squid cache.. but I think this is not correct..
need to look into it.
> - Why doesn't httpd simply replies with "not modified" to the second
> request? (Ok, may have to do with "Request-Range". I'm not to deep in
> the gory details of the protocol).
Good question. Think it should.
> According to the CHANGES document there has been a change to httpd
> 2.0.37 by Justin Erenkrantz:
>
> Fix byterange requests from returning 416 when using dynamic data
> (such as filters like mod_include).
Probably related.
> Wild speculation: httpd creates the answer "not modified" and tries to
> transfer this answer beginning with byte 985. The answer is shorter than
> 985 bytes, so this is interpreted as out of range.
If so then your Apache version is buggy..
-- Basic free Squid support provided thanks to MARA Systems AB Your source of advanced reverse proxy solutions or customized Squid solutions. http://www.marasystems.com/products/Received on Mon Jul 15 2002 - 10:17:54 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:09:15 MST