Re: Warning: header, need origin

Roy T. Fielding:
>
>> There is a real risk that frivolous use of max-age=0 will lead to
>> configurable options in user agents to switch off the staleness
>> warnings.  Must-revalidate with my proposed text provides a way around
>> this risk.
>
>No it doesn't.  What service provider would send max-age=0 when they
>know that must-validate is supposed to be stronger?

I answered this question already: Anti-social service providers
seeking higher hitcounts will find that including must-revalidate is
not good for higher hitcounts, because the associated loud warnings
which we require will scare away the public, so that they end up
getting less hits instead of more. If a sufficient percentage of the
public is using a browser which, if configured to ignore
must-revalidate, does indeed produce the loud warnings required by the
HTTP protocol, then this mechanism will not be attractive for
Anti-social service providers seeking just higher hitcounts.

>  Given that, what
>implementor will refuse to include the option to ignore must-validate
>when people counting hits include must-validate?

As said above, if we do this right, people counting hits will not want
to include must-validate, because this makes them loose more hits than
it gains them.  Thus, there would be no reason for the browser
implementer to provide a mechanism to disable the must-revalidated
warnings, because all must-revalidated warnings will indeed come from
sites that really *will* break in non-transparent mode.

>All you have done is added another name that applications will need
>to parse 

Parsing this will not require more than 3 lines of code.

>-- it does not change the economics of the situation.

It does change the economics of the situation.  In economic terms, I'm
trying to protect stateful service authors against devaluation of the
max-age=0 currency by introducing a new currency for them to use, a
currency which has more mechanisms behind it to keep it stable.

>This feature would be better accomplished my varying degrees of
>warning message rather than by different protocol elements to accomplish
>the same task.

must-revalidate is just a max-age=0 with a special warning attached.
If you want to trade must-revalidate for a Warning: header code with
the same semantics, I have no big problem.

>.....Roy

Koen.

Received on Wednesday, 10 April 1996 10:45:32 UTC