* Florian Zumbiehl wrote:
> Hi,
>
> > > The whole point is I don't do compression because of any _clients'_
> > > connections, but because of the _server's_ connection! If the
> > > server's connection usually is far slower than the client's
> > > connection (like with a server behind a V.90 modem, which would be
> > > 33.6 kb/s upstream with considerable latency), it (1) would be faster
> > > for the vast majority of clients to get an uncompressed copy from
> > > their ISP's cache rather than a compressed one from the original
> > > server and (2) even if it wasn't, it still might save traffic and
> > > thus money for the server operator.
> >
> > If I do understand correctly, you say, that not compressing the content
> > saves bandwidth? Well then turn off mod_deflate and you're done. No
> > need to change anything here, IMHO.
>
> Could you possibly point out where you think you've read that in the
> text quoted above? In case you just have a particular scenario in mind
> where at least a large part of the caches that hold a copy of the
> compressed version sooner or later get a request for the uncompressed
> version: How about considering a scenario where my suggestion could be
> sensible rather than specifically making one up where it is not?
>
> Plus, I obviously listed possible traffic reduction at the server only
> as a secondary advantage.
I did not talk about traffic, but bandwidth. I read that you're concerned
about server's bandwidth and so it would be good if the clients get
uncompressed content from the cache rather than compressed from the server.
The trick to achieve both is *to deliver uncompressed content* without
'Vary'ing the response.
nd
--
> Rätselnd, was ein Anthroposoph mit Unterwerfung zu tun hat...
^^^^^^^^^^^^
[...] Dieses Wort gibt so viele Stellen für einen Spelling Flame her, und
Du gönnst einem keine einzige. -- Jean Claude und David Kastrup in dtl
