From: "Rolf E. Sonneveld" <R(_dot_)E(_dot_)Sonneveld(_at_)sonnection(_dot_)nl>
You are creating an artificial difference between 'real world' and
I did no such thing.
From: "Rolf E. Sonneveld"
"I suspect the timeout values you use (or used) for your CBV are set
small to be useable in the real world of the Internet."
Please don't change my words. I said what I said, period.
You sure did! But I didn't change your words, Russ did. I just cited an
example of what you didn't say. I read it exactly how you meant it.
This would imply that 49% of all CBV sessions will fail due to a timeout.
What I read in Keiths words is that standards should work for 99.999
percent of all cases and we should not be satisified with success for
anywhere between 50 percent and 99.999 percent. If a new SMTP model with
built-in anti-spam protection will be designed, it would need to aim at
least to this kind of reliability (> 99.999 percent).
So in the practical world, you mean 100%?
and why 1 in 100,000? Is that good enough for you?
1 in 100,000 is way too high of a failure rate in my book and I wouldn't
presume a new SMTP model to begin with a broken concept regardless of how
small it may be in the final analysis.
Anyway, you were the 1 in 1.1 million for our system. Now that is what I
Hector Santos, Santronics Software, Inc.