Re: [tcpm] Some comments on 2581bis

> I think it would be helpful to describe possible mechanisms (such as
> the one above). If not, chances are that people is not going to
> implement it, or they may get the limit wrong. The doc already some
> includes text about ack-division attacks, so IMHO I don't think
> describing a mechanism for enforcing limits on dupacks would be out of
> scope.

The hesitation here is that we're making things up for inclusion in a
document that we are trying to progress to draft standard.  I am a
little uncomfortable with that.  The other things in the document have
some more long-term understanding and study behind them.

I don't mind the document saying "here is a case where you might want to
be more conservative than allowed".  But, if we translate that into one
or more mechanisms we could easily get bit.  I have done this "it is
such an easy tweak, how could it be wrong?" business enough times over
the years to know that the interactions are tricky and subtle and it
takes work to understand the implications of these things.  And,
sometimes the "easy, for sure gotta be right" changes are absolutely
wrong.  I am just feeling like we don't want to do that in this
document. 

Just my opinion ...

allman