7/29/94  -- an update to emily.*

	1. A bug in displaying the scaled value in the information dialog 
	   when using log scaling was fixed.  Previously, it displayed NaN 
	   if the input was negative.  This only affected
	   the info. dialog -- the display matrix values were calculated
	   correctly.

	2. The matrix addressing has been changed so that the upper left
	   hand corner is (1,1) and the lower right corner is (NR,NC). 
	   Previously, the upper left corner was (0,0) and the lower right
	   corner was (NR-1,NC-1). 

	3. The info dialog will now be updated using drag operations (instead
	   just once per click).  In other words, once you've popped up the
	   info dialog, as long as you keep the mouse button down, the info
	   dialog will be updated as you move the mouse.  As before, info 
	   dialog updates won't occur during a zoom operation.

	4. a note has been added to section 4.4 of the Users Guide attempting
	   to clarify when magnitudes (abs. values) of input are used for
	   scaling.


8/31/94 and 9/1/94 -- bug fixes

	1. Incorrect input min/max values were displayed when the input was 
	   a Matlab sparse matrix.

	2. Better handling of an incorrectly set DISPLAY variable.

	3. Incorrect input min coord values reported if all non-zero values were the 
	   same in a zoom window.

9/14/94 -- minor improvement and bug fixes

	1. Incorrect input min/max values were displayed when the input was 
	   a Matlab dense matrix.
	
	2. standalone version now takes another option: if "-" is given as
	   the file name, emily reads from standard input to allow use
	   at the tail end of a pipe.

	   Example:    Suppose you have 2 executables, big_solver and reorder,
		that do as their names suggest, solve a linear system and 
		then does some kind of reordering and you'd like to see
		the reordered output using emily.  A pipe might look like:

		big_solver | reorder | emily -

10/6/94 -- minor improvements

	1. Finally (I hope) the max/min problems are fixed.  

	2. Minor change to handle matlab 4.2

	3. Now have matlab versions of emily and hbin for the SGIs
	   emily appears to have some bugs in it, which I hope to
	   work out (or find out that it's an compilation environment
	   problem.)

10/30/94 -- fixed a bug with hbin.  In some cases, it would fail to read
            in a matrix, if the last row of the input file was not the same
	    length as the rest.  This bug is fixed.

12/14/94 -- Installed SUN 4 emily and hbin mex files that were compiled at
	    the University of Minnesota -- Thank you very much for letting
	    me steal a few cycles.

1/27/95  -- Added a print option to the file menu, which prints PostScript to either
            a file or prints it using a command.  Also fixed a bug with hbin reading double
	    precision input.

1/30/95 -- Added executables for DEC MIPS (emily.dec) and Alpha (emily.alpha) based machines.
	   Matlab executables for both architectures are available on request (write me
	   at tloos@cs.indiana.edu and I'll get them for you)

2/24/95 -- Installed a couple of demos for emily under matlab.  emily_demo.m needs the demo.mat
           file and mydelsqdemo.m is a modified version of MathWorks delsqdemo with a couple
           of emily calls installed.  mydelsqdemo depends on mydsqshow.m


3/22/96 -- UPDATES ONLY MADE TO SGI and STATIC.SUN version (so far)
	support partitioning lines being drawn on the matrix.  By supplying
an optional second argument to the standalone version only, a "partitioning
vector" file can be read in and grid lines can be drawn.   A partitioning
vector file can be of one of 3 formats:

	1. a sorted Chaco output file (emily doesn't handle reorderings)
	2. a block start file of the following form:
	   ptn starting row 1
	   ptn starting row 2
	   ...
	   ptn starting row nptns+1
	 
        Typically, the first row is 1, etc.  For a 4-way equipartition of
	a 1000 row matrix, the block start file would be:
	   1
	   251
	   501
	   751
           1001

	This puts rows 1-250 in the first partition set, 251-500 in the
	second etc.

	3. The same as 2, but with the number of partition sets as the 
           first entry of the file, so this would change the example above
	   to be:
	   4
	   1
	   251
	   501
	   751
           1001
	
	Also, a few small bug fixes were made.

	Let me know if any other architectures need support.  The matlab
        version doesn't support this feature, so it wasn't updated.

	Send any bug reports or suggestions to: tloos@cs.indiana.edu