PJACKLAM / Math-String-Charset-Wordlist-0.09 / CHANGES 
2017-02-02 0.09 pjacklam

 * Trim whitespace.

 * POD nits (CPAN RT #51825).

 * POD formatting.

 * The NEW file now refers to the CHANGES file.

 * Revert to ExtUtils::MakeMaker since use Module::Install is discouraged.
   Change license from "perl" to "perl_5", which is the recommended for Perl 5.

 * More documentation in POD about BUGS and SUPPORT.

 * Update README file.

 * Remove INSTALL file. The information in that file is now in the README file.

 * Required version of Test::More is 0.47, not 0.62.

 * Required version of Math::String is 1.22, not 1.28.

 * Required version of Math::BigInt is 1.74, not 1.87.

 * Remove Math::BigFloat from list of requirements, since this is bundled with
   Math::BigInt in the Math-BigInt distribution.

2008-04-03 0.08 Tels 181 tests

 * modernize package (bundle Module::Install etc.)

 * add POD tests

 * we don't need to have Exporter in @ISA

Complete version history
========================

##############################################################################
Math::String::Charset::Wordlist

2003-02-06 v0.01 Tels
  * first version

2003-04-14 v0.02 Tels
  * fix for scale not beeing a legal parameter to new()
  * added method copy() (with special case for Tie::File object)

2003-04-27 v0.03 Tels  81 tests
  * added: file(), class()
  * fixed: str2num: use the "normal" sorting of the wordlist (like
 	   "sort -u" produces), instead of "shortest strings first"
  * fixed: ones(), start(), end(), char()
  * made it's own distribution with XS code to replace File::Tie
    _file(), _free(), _offset() and _record()
  * fixed: str2num(): use $self->{_len}->numify() instead of $self->count()
           This makes the search _much_ faster since it no longer uses
	   BigInt math (e.g. 1800/s vs. 70/s)

2003-04-27 v0.04 Tels  152 tests
  * speedup str2num: just set $leftstr/$rightstr from $middlestr instead of
    fetching them over and over. (2470/s instead of 1800/s)
  * _offset(): return undef when a negative record number is passed
  * offset(-$n): return offset counting from last record backwards
  * _record(): increase limit from 1024 to 8192 bytes. (still, fixed limit is
    baaad...)
  * _record(): remove 0x0A and 0x0D from record end (to cater for win32 files)
  * _file() reads in the file blockwise (8 KByte) instead of bytewise, this
    makes if _much_ faster again:

   # perl time.pl big2.lst
    Math::String::Charset::Wordlist v0.03
    new() took 0.549798965454102 seconds.
    contains 913954 words.
    FETCHALL took 0.000316023826599121 seconds.

    perl -I/lib -I/blib/arch time.pl big2.lst
    Math::String::Charset::Wordlist v0.04
    new() took 0.0736250877380371 seconds.
    contains 913954 words.
    FETCHALL took 0.000316977500915527 seconds.

2003-04-28 v0.05 Tels  178 tests
  * if record len was 0, _record() bombed (buffer underflow)
  * dictionaries with more than 8192 bytes did not work, due to bug in offset
    reading
  * num2str() did not work for negative numbers (abs() missing)
  * more tests
  * we need Test::Simple, so list it in Makefile.PL

2004-08-13 v0.06 Tels  179 tests
  * override chars() from Charset.pm and return 1
  * minlen()/maxlen() can be inherited
  * adjust prerequisites
  * fix warnings about unused vars in XS code

2004-11-20 v0.07 Tels  179 tests
  * make to work with Math::String v1.25