version 0.16  10/30/2012

    Small but important fix: modified the control in "calc_trans()"
    inside atpmatch.c so that if one tentative match fails, the
    code does NOT immediately quit; now, it will go on to other
    possible matches.  More robust, but also can now report
    less-than-optimal matches.  May require more tuning to find the
    optimal match.


version 0.15  08/17/2012

    Big change, removing the brute-force method of Valdes et al. 
    (PASP 107, 1119, 1995) for the more clever and robust technique
    of Tabur (PASJ 24, 189, 2007).  Tried to keep input/output/arguments
    the same, with very minor exceptions to input args.

	
version 0.14  11/04/2011

    Added "rotangle=" and "rottol=" arguments, so that the user can 
    now specify a permitted range of rotation angles between the
    two coordinate systems.

version 0.13  11/30/2010

    Added a Unix-style man page which is extracted by hand (ugh) from
    the .html manual.

    Added an "install" target for Makefile

version 0.12  11/22/2010

    Fixed the description of the "outfile=" argument in the USAGE section
    of "match.html".
   
    Added a printf statement in "make_vote_matrix()" in atpmatch.c.

version 0.11  6/26/2010

    Changed output TRANS elements format from %15.9f to %15.9e; should
    increase accuracy of subsequent calculations when values are small.

version 0.10  6/11/2008

    Small bug fix in "apply_match.c"

version 0.9   2/17/2007

    Added notes in documentation to warn against very large ranges
    in the input coordinate systems.  If the uncertainty in position
    of each item is more than about 50 units, matching will fail.
    See #define AT_MATCH_MAXDIST in atpmatch.h

version 0.8  10/25/2003

    Added an option to the "project_coords" program at the request of
    John Blakeslee and the ACS team.  The new "asec" option causes
    the program to print the (xi, eta) coords of each star in arcsec
    away from the projection point, rather than in radians.


version 0.7  5/19/2003

    Small fix to "apply_match" program: now handle coordinates near the
    celestial poles properly.
     -- MWR 5/19/2003


    Small bug fixes:
        a. match.c      if "reset_A_coords()" passed empty lists,
                                it no longer fails assertion, but prints
                                warning message and continues
        b. atpmatch.c   added similar checks in "atCalcRMS" and "atFindMedtf"
                                to deal sensibly with lists of zero stars
                                instead of failing assertions.
     -- MWR 10/23/2002


    Fixed "misc.c" so that "read_data_file" can now really handle data files
    with up to 20 columns; a bug prevented any data in columns >= 10 from
    being read properly.
     -- MWR 8/17/2002



version 0.6  6/19/2002

    Enhancement, code supplied by John Blakeslee
             new routine, atCalcRMS, calculates "sx" and "sy" for
             all matching items in each direction, using 1 iteration
             of 3-sigma clipping.  Adds two new fields to end of 
             the TRANS output line


version 0.5  12/31/2001

    Enhancements (thanks to John Blakeslee)
	      a. 'identity' option forces match with no scale change or rotation,
            with no translation (by default), or a given translation
            (if 'xsh=' and 'ysh=' arguments given)
         b. 'intrans' option allows user to specify exactly an initial
            TRANS structure
         c. after initial TRANS is discovered (or given by user), the
            program will always run 'atRecalcTrans' using only the 
            matching items

    Fixed several small bugs; thanks to John Blakeslee and Andrew
    Bennett.

    Added more test cases to 'selftest.pl'.


version 0.4  12/14/2001

    Fixed small bug in "project_coords.c" which limited number of
	 columns in data files to 10 (instead of the stated 20).

	 Modified "write_array" in atpmatch.c so that output files
	 "matched.mtA", "matched.mtB", "matched.unA", "matched.unB"
	 have coordinates printed with format 13.7f, instead of 12.6f.
	 When coordinates are radians, this provides granularity of
	 0.02 arcsec, versus 0.2 arcsec with old format.  The extra
	 precision may occasionally be warranted.

	 Added new elements to the TRANS structure, which provide a summary
    of the quality of the fit.  

    New option to produce a second line of output, the MEDTF structure,
    in case the two files are known to differ only by a pure translation.

    New "min_scale" and "max_scale" arguments are permitted, as well as
    the old "scale" argument.  



version 0.3   1/21/2001

    Added new command-line arguments "max_iter" and "halt_sigma",
    which allow user to control the number of iterations of
    the main program loop.  This allows the user to terminate
    the search for solutions when the residuals drop to a particular
    level, for example.

    Fixed small bug in "selftest.pl" script which caused the test
    of "id1" and "id2" options to fail.    MWR 10/29/2000

version 0.2   7/30/2000

    Added "id1=" and "id2=" options so that user-assigned ID numbers may
    optionally be used throughout program and kept in output files.


version 0.1   6/14/2000

    Initial release of the package.  Includes the workhorse program
    "match", plus a pre-processor (project_coords) and a post-processor
    (apply_match).  

    It would be nice to modify the pre-processor to handle (RA, Dec)
    coordinates expressed either in decimal degrees (in current version),
    or in sexigesimal notation.

    The regression test "selftest.pl" tests only the "match" program.
    There are no regression tests for "project_coords" or "apply_match".



