Some iprop 0.7.2 questions.

["Henry B. Hotz" <hotz@jpl.nasa.gov>]

Shouldn't kadm5_log_previus() in lib/kadm5/ check if it's backspacing  
before the beginning of an empty (or nonexistent) file and give a  
warning?

Shouldn't replay_log have a --reset option that erases the log file  
after the replay (the way it's done at the end of receive_everything 
() in ipropd-slave.c is probably good).

Does it ever make sense to do a replay_log on a master?  Maybe if  
you're doing slave -> master promotion, but otherwise?

I've seen some reverse-channel effects where an ipropd-slave can  
cause changes in the master's log file, and I need to trace those  
down now.
------------------------------------------------------------------------ 
----
The opinions expressed in this message are mine,
not those of Caltech, JPL, NASA, or the US Government.
Henry.B.Hotz@jpl.nasa.gov, or hbhotz@oxy.edu