Re: minimum-cache time?+

Hi,

Christian Balzer wrote:

> 1. Objects with short expire values, which can be quite reasonable in
> fast changing page. For these I'd like to see an option which changes
> the forcible reload into an IMS request, so if nothing changed, nothing
> has to be fetched. ^_^

Sounds reasonable. About the MS server, though, I can imagine that
they've declared their sites "Best cached with Catapult/IIS",
which sends a secret MIME header to the server to make it behave
rationally...

> Of course this fails if no Last Modified headers are available, so we
> also need...
 
> 2. If there are sites which make caching really hard on us, even if
> they don't change (all) their date so frequently, we need some ways to
> define sensible TTL/Refresh values for them, which should be fairly
> easy with the current pattern mechanism.

Like, one hour? But then we're working out a technical solution which
is a design/politics problem. The web designers of the sites in question
think everyone has a lot of cheap bandwidth. Basically, these web
sites are designed for a paradigm where your browser can communicate
quickly with the server. But for those of us who don't have that
luxury...

> It might be a violation, but so is this foul game for hits they play.
> Alas your suggestion lacks a method to define how long you want to
> chache these... ^_-

You don't need one; my proposed "must_cache" directive would mean
that Squid will calculate the object's lifetime according to the
other rules.
>

> Hehehehehe, thank you!

I hope it works. In solving the caching problem, we can't just work
on the cache end--the server should also cooperate. Hopefully, the
commercial server vendors and web site developers listen up.

-- 
miguel a.l. paraz  <map@iphil.net>  http://www.iphil.net/~map/  PGP: 0x43F0D011
iphil communications:  isp/intranet design and implementation, makati city, ph