[ns] TwoState Error Model?

[Anaking <anaking@sina.com>]

Hi, all,

I search mailing list archive, but can't get a specific conclusion. I need 
confirmations.
First, ErrorModel/TwoStateMarkov has the same function as
ErrorModel/Expo, since the transition code has been commented and transition
play no role. The error rate is exponential distribution and can be specified
but mean good/bad length can't be specified. The mean good/bad length is always
between [0,1], does it means the mean value is 0.5?

Second, the improved multistate version of Pan Jianping can be used as
TwoStateMarkov model, right? And the transition matrix is useless for mean
lengths can account for the transition?

Third, when 
set em [new ErrorModule Fid]
$link errormodule $em
"warning: classifier _o93 unknown flow s:0, d:256, fid:0, bucket:0
_o93: no target for slot -1" exits.
what cause thi error.

Thanks for your help.