Re: Why MD5 Headers are Imperative

At 11:09 PM +0000 8/17/06, Eric J. Bowman wrote:
>  >
> > -1. This is vast overkill, as shown by other responses on this thread.
>
> I'm sorry, Paul, but this is why I was afraid to be so presumptuous as to
> start a Pace.  I was afraid that it would immediately get shot down in a
> flurry of "-1's" without anyone stopping to consider the underlying problem
> it brings up regardless of what they think of my posting style, list
> seniority, solution or other irrelevancies.

The main purpose of Paces is to give an entire argument, including 
the proposed text for the spec, in a single place. This gives people 
a way to decide whether or not your reasoning about the underlying 
problem is valid, and whether your specific proposed solution matches 
the scenario is correct.

This WG has seen cases where an underlying problem brought up in a 
Pace was agreed to, but the proposed solution was completely change. 
The WG has also seen Paces where the underlying problem was expanded 
because people saw that the problem was, after reflection, bigger 
than the Pace author saw. The WG has seen Paces pretty much 
rubber-stamped as "yup, looks right to me".

This WG has also gone down rat-holes where the author of a message on 
a thread keeps changing the problem, or the proposed solution. This 
is a Bad Thing.

> What other responses, aside from James agreeing with this message, claim that
> this is vast overkill?

None, and I didn't say they did. I expressed my opinion.

> Julian challenged my assumptions, started to see that
> I may indeed have a point, and agreed to disagree

I read Julian's comments in a different light that you have.

> -- you're my next
> commenter,

No, there were others.

>  would you care to elaborate on your position as to my problem as
> pertains to its technical merit?

Sure. I did not hear anyone saying that they thought the problem was 
as bad as you thought, and others have said they don't understand 
what you are saying. Until there is strong agreement that there is a 
serious problem, we should not consider a solution such as that would 
be difficult to implement. "Change HTTP" seems pretty darn difficult.