rejsizes patch.

[Russ Allbery]

Dan Merillat <harik at chaos.ao.net> writes:

> Thanks to Katsuhiro for the oneliner, it's now a lot more accurate.
> Without those 1gig articles sprinkled liberally through my logs,
> I ended up with a much more reasonable 77% accept rate/79% size.

> Going to do some more tweaking, like adding reject sizes to the
> per-hour list so I can get an idea of true bandwidth.

> Anyone interested in merging this in?

Yup, sounds like it should go into 2.4.0 to me.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>