Release Notes:  Earthworm Version 2.4


MODIFICATIONS TO EXISTING MODULES:
**********************************


EQVERIFY: One of the sausage routines. 

1) Eqverify v2.31 and v2.4 have different tests on the temporal
   distribution of P-arrival times:

v2.31 uses function test_ptimes() which is based on Dave Oppenheimer's
suggestions.  test_ptimes studies only picks arriving within "MaxDeltaT"
seconds of the first-arriving pick.  It uses the median delta-t with 
respect to the median arrival time in judging the P-arrival distribution.
This test may label some small earthquakes in dense parts of the network 
as noise.  It also allows some events that contain 2 groups of glitch 
picks to slip through as earthquakes.

v2.4 uses function test_pgroup() to look at the relative arrival times of 
the picks to find groups of picks that belong to glitches.  In this test, all 
arrivals within "MaxDeltaT" seconds (default = 30) of the first-arriving 
pick are examined to see if they belong to a glitch as defined by default 
settings (at least 4 picks within 0.03 seconds) or by the "define_glitch" 
command.  If 50% or more of all picks belong to glitches, the event is 
declared noise.  If 30-50% of the picks are glitch-picks, the event weight 
is incremented by 1.
For comparison, v2.4 still performs the test_ptimes() and prints a line
for it in the log file, but it doesn't actually use its results.

2) New config-file commands for v2.4:
The following commands were added to allow the user to switch each
noise test on or off (the tests are described in some detail in 
.../user_doc/eqverify.commands).  These are required commands.
	test_codawt	<switch>
	test_freefitrms <switch>
	test_pgroup 	<switch>
	test_slopevsmag <switch>

This command was added to set up the definition of a glitch for the
test_pgroup function.  It's an optional command that changes default
settings.
 	define_glitch <npick> <nsec>




LOGIT:

1) Bug fix on OS/2 Version: Time handling corrected so that UTC time
will be correctly calculated and reported in the log files.
	


BINDER:

1)  Bug fix in grid.c  4/8/96
Previously associated, then culled picks sometimes get re-stacked
into new events.  The bug was that the phase label from the old event
was carried into the new stack when in fact all picks in a new stack
should be labeled as "P" (example: if a pick was associated as "Pn" in
a previous event, it was also called "Pn" in the newly-stacked event).
This mis-labeling of phases sometimes caused location problems.
A one-line addition to grid_stack() in grid.c forces all picks in a
newly stacked event to be labeled "P".

2)  Bug fix in hyp.c  4/24/96
In hyp_pau(), these was a potential divide-by-zero problem in the
calculation of ravg (the average epicentral distance).  Fixed.
 
3)  New feature in grid.c
Binder now has the ability to ignore glitches in its stacking phase.
The following command was added to configure this feature:
 
define_glitch <npick> <nsec>            grid_com                stack
        Sets up the definition of a glitch for filtering arrivals before
        stacking them.  A glitch is defined as at least <npick> arrivals
        within <nsec> seconds.  Picks that belong to such a glitch are NOT
        used in the stacking process.  If you want to turn off pre-stack
        glitch filtering (ie, to make binder stack every pick), set <npick>
        to zero (example "define_glitch 0 0.0").
        Default:  define_glitch 4 0.03          Calnet:  define_glitch 4 0.03
 


MENLO_REPORT:  

Modified to log each event copied to remote system.


CHANGES TO CONFIGURATION FILES:
*******************************

binder.d:	See above under binder 3.

eqverify.d:	See above under eqverify 2.


KNOWN BUGS:
***********

Hypoinverse:

Hypoinverse will occasionally die or hang on an event.  The Earthworm
recognizes that this has happened and sends an error message.  Once this
happens, the system can no longer process earthquakes.

It is recommended that a pager message go out when this error occurs and
that the responder reboot the OS/2 box.


Adsend:

Picker does not pass on adsend module heartbeats, maybe no error messages either.


Menlo_report:

Logs twice when it sends arc file to remote system.


Statmgr:

The statmgr 'stopping' message does not always get out to pageits and email.


Transport Layer:

If the input buffer length is set too small by a program, the transport
layer no longer returns an error message.  Instead it hangs the system
under both solaris and os2.


MODULES IN PROGRESS:
********************

	Export - version that forks under Solaris.

	Import_generic

	Export_generic

	Getdst

	Lptrig

	Mteltrg

	Picker_exp

	Wave_server

	Wave_client



