Re: Ted Hardie's comments on draft-mogul-http-hit-metering-01.txt


The audience for a proposed standard (which is what you want
this draft to become) is not just the working group. I think
that proposals are stronger if they describe the problem
they're attempting to solve before they outline the solution,
so I endorse Ted's suggestion that you fix the ordering.

I also found your informal description about critical path
scheduling to be enlightening and wish you would include it
(nearly as you wrote it in the message) in the draft. I think
it's OK if you include tutorial material about what a 
"critical path" is in the course of analyzing network
protocol latency. 

Everyone's unwillingness to understand Koen's drafts on variants
may result in us just not making progress. Won't you please
give it a try?

Also, I thought we had a clear agreement that you would reduce
the applicability claims in the draft to the point where they
were no longer controversial: that is, that hit-metering
was defined as just a way to do hit metering for those cases
where this kind of hit metering was useful, and to stay
away from endorsing it as "good" or even "adequate." It doesn't
really help, anyway.

"Based on the following analysis, this design will be
 useful for user-counting under some circumstances, while
fitting within the constraints of what is feasible on
the current Internet."

I think you will do yourself and the IESG a service by identifying
in the document itself the ways in which it is controversial,
and at least mentioning the alternatives that have been proposed
("statistical sampling") even though not explored.

I guess I'm a little disappointed that we still have a draft
that is unnecessarily controversial.


Received on Tuesday, 18 February 1997 17:19:53 UTC