5.2.0 --> 5.2.1 (Ivano Primi <ivprimi(at)libero(dot)it>) :

*   Corrected a bug which prevented Numdiff to be built on MacOS X(R).
    Thanks to Antoine Dechaume for the report.

5.0.0 --> 5.2.0 (Ivano Primi <ivprimi(at)libero(dot)it>) :

*   Numdiff uses now the GNU Multiple Precision Arithmetic Library
    (also called GNU MP or GMP) if this library is available 
    at build-time.
    The old internal support for multiple precision arithmetic
    is a fall-back in case GNU MP is absent.
    However it is possible to use the internal support for 
    multiple precision arithmetic even when GNU MP is available:
    it is sufficient to pass the option --disable-gmp
    (or --enable-gmp=no) to the  configure  script
    before building the program.
    After building and installing Numdiff,
    it is possible to know whether it has been
    linked against GNU MP or not just by calling the program
    with the option -v. In case of link against GMP
    the program tells also which version of
    GNU MP it uses to perform its computations.

5.0.0 beta4 --> 5.0.0 (Ivano Primi <ivprimi(at)libero(dot)it>) :

*   The filter has been improved, now it gets no more confused if
    the same value is represented in scientific notation in one file
    and in common decimal notation in the other file.

*   The Makefile.in (and consequently the Makefile) has
    been modified to conform to the GNU Coding Standards.

*   Now numdiff and ndselect accept long options too
    (for instance --separator='\n\t %' instead of -s '\n\t %').

*   The option -r accepts now an additional parameter which tells 
    to compute the relative differences always with respect 
    to the values in one file, instead of using the classic 
    formula for the relative error. 

*   The argument of the option -f has been made optional.

*   Documentation, i18n and italian l10n have been updated.

5.0.0 beta1 --> 5.0.0 beta4 (Ivano Primi <ivprimi(at)libero(dot)it>) : 

*   Added additional information to the report printed by
    the option -S.

*   Now the program can accept the options with argument

    -z INT, -Z INT, -X INT

    as short equivalent forms of 

    -z 1:INT -z 2:INT, 
    -Z 1:INT -Z 2:INT  and
    -X 1:INT -X 2:INT  respectively.

*   The option  -R  has been removed since not useful.
	
*   The option  -s  has now an extended syntax allowing to specify
different sets of field delimiters for the two files to compare.

*   Adoption of the version 1.3 of the GNU Free Documentation License
for the TeXinfo documentation.
	
*   Now the test programs are not anymore distributed together with
    the source code of Numdiff.    
	
4.2.0 --> 5.0.0 beta1 (Ivano Primi <ivprimi(at)libero(dot)it>) : 

*   Removed the support for double, long double 
    and high precision arithmetic. Now only multiple
    precision arithmetic is available (but it is not a limitation :) ).

*   Removed the option -L, since the tool ndselect made it obsolete.

*   Implemented a synchronization procedure to detect insertions/deletions
    of lines. To accomplish this, source code from GNU diff (diffutils 2.8.1)
    has been used (with the necessary modifications).
    Added the options -f, -z, -Z, -R, -m, -H, -T.

*   Removed the option -F, -X is now used in place of it.	

*   Added new entries to the help which explain how to use the new options.

*   The output format has been slightly changed.

*   Makefile and configuration scripts have been changed, since
    the code has become much more complex than it was.

*   Added more tests to check that the new features are correctly working.
	
*   Change of the license: from GPL version 2 or later to GPL version 3 or later.
    The copyright statement of my files has been consequently updated.
    Also the message shown by 'numdiff -v'.

4.0.0 --> 4.2.0 (Ivano Primi <ivprimi(at)libero(dot)it>) : 

*   Changed the way the options -V and -b interact when they are both set.

*   Added new tests.

*   Now, when the options -v and -h are both set, Numdiff displays before
    the information about version, Copyright and NO-Warranty and then
    the help message.

*   The Copyright has been updated.

*   Added the tools `ndfilter' and `ndselect'.

*   i18n and italian l10n updated.  

*   The documentation has been consequently updated.

3.0.2 --> 4.0.0 (Ivano Primi <ivprimi(at)libero(dot)it>) : 

*   New output format. 
    At the same time the code printing the output has been cleaned.

*   Added the -V option for the "verbose" mode.

*   Better handling of the last line of a file.

*   i18n, l10n and documentation have been consequently
    updated.

3.0.1 --> 3.0.2 (Ivano Primi <ivprimi(at)libero(dot)it>) : 

*   The code of the function Cabs() has been improved
    for both double and long-double precision.

*   Fixed a compilation error which occurred when one
    chose to activate the double, the long-double or the high
    precision instead of the multiple precision arithmetic.
    This compilation error was introduced just before releasing
    the version 3.0 of Numdiff .

3.0   --> 3.0.1 (Ivano Primi <ivprimi(at)libero(dot)it>) : 

*   The code of the function Cabs() has been modified
    in order to speed up the execution when working with
    real numbers (the change concerned only the code
    for multiple precision arithmetic).

3.0 (Ivano Primi <ivprimi(at)libero(dot)it>) : 

*   First public release
