---- what's new in ReDIF-perl v 2.23 ---------------------------------------

  - fixed a very small problem in ReDIF/init.pm which caused wrong
    start-up behaviour sometimes.

---- what's new in ReDIF-perl v 2.22 ---------------------------------------

  - redif.spec: changed case-normalization of the Person template
    handles to make it compatible with established practice

  - spec/redif.spec file would not anymore appear in the packages to
    minimize confusion between three redif.spec files present in the
    package (the third one is a test's private copy in t/ dir).  Use
    lib/ReDIF/redif.spec if you need the actual specification.

  - a "closing an already closed template" complain (Carp::cluck)
    removed, bacause it was appearing when a bad-started template was
    read right after some valid ReDIF...

---- what's new in ReDIF-perl v 2.21 ---------------------------------------

  - a very serious memory leak, discovered by Sune Karlsson, fixed.
    (Although I still don't know the reason why an eval operator
    leaked.  Probably it shouldn't have and it may be a bug of perl.)
    As a side-effect, performance improved greatly, roughly twice at
    my estimation.  But still performance is lower than that of the
    older before-unicode version (e.g. 2.16)

  - fixed minor problems in rr.pm which caused complains in "perl -w"
    mode

  - added redif_open_dir() and redif_open_dir_recursive() functions to
    the ReDIF::Parser.  That is to ease porting from now deprecated
    rr.pm-based applications.

  - fixed a problem in ReDIF::init module, which didn't allow to
    actually use install-time defined ReDIF home directory setting.
    It was always trying to use current directory as home, unless
    specified with command-line option or with an environment
    variable.

  - also now ReDIF home is NOT anymore NECESSARY to use ReDIF::Parser
    or rech.  Since redif.spec is installed with all .pm files, we do
    not actually need the home for many operations.  So if your script
    used ReDIF home dir, provided by ReDIF::init, you probably should add
    a check for it (look in rech for an example).

---- what's new in ReDIF-perl v 2.20 ---------------------------------------

  - rech output improved/fixed

---- what's new in ReDIF-perl v 2.19 ---------------------------------------

  - new templates validation rule: if an error happens while checking
    value of an attribute which is not required, the error is turned 
    into a warning

  - now rech always quotes the original data text on errors & warnings
    (before that it was close to original but reformated).

  - fixed processing of "X-" attributes, including cluster-level:
    (e.g. author-x-haircolor: red)

  - removed huge flow of warnings generated in "perl -w" mode.  that may be
    incomplete.  please let me know if you come accross any unfixed causes

  - unicode (UTF-8) support.  see README.unicode for more info

  - completely new parser interface module ReDIF::Parser with new
    capabilities (see ReDIF::Parser manpage).  use it instead of
    outdated and deprecated rr.pm!

  - changed redif.spec format: more readable, easier to understand.
    among the rest, it now has version parameter.

  - new redif.spec features: RePEc handles case-correction and some other

  - now continuation lines in ReDIF templates doesn't have to start
    with a whitespace, it would be enough that they don't start with 
    an attribute-like construct: /^[a-zA-Z\-]+:/ (so called
    "sloppy-lines")

  - another potentially problematic change: now even if using rr.pm,
    the attribute values don't have some formatting it used to have.
    Line-break characters in the multi-line attribute values are not
    eliminited anymore.  They used to be joined into a single line.

  - changed some options of rech and rere, see their respective
    manpages for details.  rech now doesn't need a configuration file
    and has new (improved) output format    

  - ReDIF::init's initialize() is now silent by default

  - added lots of tests for automated and simplified code quality control

  - added/updated more documentation bits and pieces in pod (manpages)
    and in READMEs

  - almost every piece of old code rewritten, became more readable,
    manageable, more logically structured into modules

