Re: gzip encoding weirdness.

On Sun, 2002-08-25 at 23:36, Adrian Chadd wrote:

> [localhost:~] adrian% telnet www.enetica.com.au 80
> Trying 203.17.36.111...
> Connected to www.enetica.com.au.
> Escape character is '^]'.
> HEAD / HTTP/1.1
> Host: www.enetica.com.au
>
> HTTP/1.0 200 OK
> Date: Fri, 23 Aug 2002 10:21:13 GMT
> Server: Apache/1.3.17 (Unix) PHP/4.1.2 mod_gzip/1.3.19.1a mod_ssl/2.8.0 OpenSSL/0.9.6
> Last-Modified: Fri, 16 Aug 2002 08:09:37 GMT
> ETag: "163206-294b-3d5cb341"
> Accept-Ranges: bytes
> Content-Type: text/html
> Content-Encoding: gzip
> Content-Length: 2661
> Age: 1427
> X-Cache: HIT from <foo>
> X-Cache-Lookup: HIT from <foo>
> Connection: keep-alive

The origin server SHOULD (RFC 2616, section 14.44) have provided a
Vary: Accept-Encoding
header (assuming that Accept-Encoding is the header used to make the
decision. See RFC 2616 section 14.3).

> Can someone HTTP/1.0 and HTTP/1.1 cluey explain whether this is valid or
> invalid, please?

it's valid operation on squid's behalf, invalid on Apache's. We've had a
couple of bugs on this:

193
378

Someone needs to hit the apache mod-gzip folk with a cluebat.

Rob