Re: SHOULD-level requirements in p6-caching

On Fri, Apr 08, 2011 at 11:14:07PM -0700, David Morris wrote:
> 
> 
> On Sat, 9 Apr 2011, Willy Tarreau wrote:
> 
> > On Fri, Apr 08, 2011 at 05:33:42PM +0000, Poul-Henning Kamp wrote:
> > > In message <Pine.LNX.4.64.1104080808000.18147@egate.xpasc.com>, David Morris wr
> > > ites:
> > > 
> > > >My expectation is that current sites which use cookies to control
> > > >content already use other caching controls to prevent reuse of
> > > >unsharabale content.
> > > 
> > > IMO an unwarranted assumption.
> > > 
> > > As author of a major server-side cache-software, I see a lot of
> > > server side people with absolutely no clue to caching and its
> > > interaction with cookies, vary or for that matter cache-control
> > > headers.
> > 
> > I agree with you, I've had the same experience. People develop
> > applications and install them in a hosting infrastructure they
> > don't necessarily understand. The hosting infrastructure uses
> > caches and the application people complain that the caches are
> > "abusive" and distribute session cookies to end users...
> > 
> > I even had to develop specific options in haproxy to test for
> > response cachability combined with set-cookie headers and be
> > able to block them instead of letting them leak into caches.
> > 
> > For me it's a proof that there are many application people who
> > have no clue about caching and who sometimes don't want to know
> > how it works.
> 
> And changing the spec would help that how? If feature that already
> exist aren't being used, why does another feature help, in particular
> in the face of the normal propigation delay between specifiction
> and sufficient deployment.

I was not discussing a change of spec, rather the point that was made
that people who use cookies know how to use a cache. That assumption
cannot be made, so if it's used to justify a spec change, it's hasardous.

Regards,
Willy

Received on Saturday, 9 April 2011 06:28:47 UTC