Dave and I have exchanged some email about Md and Ml and
how we want to proceed in the new system.  We would like to
explore the approach of using the output from eqprelim
(events that have more than 25 picks) to initiate a cmag
computation to get a rapid magnitude estimate out.

We've gone back and forth on this a bit - my thought was
to look at pulling events out faster than eqproc does now.
But I think this idea has merit and I'd like to explore it.

In this approach, events from eqprelim would trigger cmag or a
cmag relative running with a different configuration.  If 
the cmag Ml/Me meets some criteria (>~3.5,minchannels,etc), cmag
would report to the database.  We would want alarming actions such
as qdds and some level of notification to take place.  If the cmag
Ml/Me failed to meet these criteria, the location only would be
reported to the database and to QDDS.  When a final solution is
available, it would trigger cmag and new magnitudes (associated with
the updated location) would be generated and stuffed in the database.

At this point, we would only trigger cmag.  No other higher level
of processing would be triggered by this (no MT, no ampgen, no ShakeMap,
etc), although it might be worthwhile to consider that possibly some other
modules might get triggered in the future.

I think we talked over the idea of how to handle the eqprelim and
eqproc feeds in our last meeting (and how to track and handle deletes); 
this is a more concrete plan of what we would like to do with it, now 
that we have some idea of what rules would make reporting the Ml/Me 
feasible.