==[ v0.1, Tue Jan  4 02:24:40 CET 2005 ]==
- First release (internal)

==[ v0.1.1, Thu Jan  6 17:13:56 CET 2005 ]==
- Public release
- no code changes, only README/etc updates

==[ v1.0beta1, Mon Jun 13 02:26:16 CEST 2005 ]==
- Moved all configurable parts (except DEBUGMODE) to the configfile.
  Besides making it much easier to change configuration items, this
  also makes it useful to have a precompiled windows version now.
- Added ::except-hosts, which is especially needed for CGI:IRC
  gateways, suggested by RyaN and Niklas.
- Improved default regexes to include more digits names. I explicitly
  did not include 4 digit names however because some users might have
  nicks like 'Girl1984' etc. Suggested by RyaN.
- I didn't notice this earlier on, but the score values for the regexes
  were simply ignored and always +1... This has now been fixed. As a
  side effect this means the regex scores have much more impact so I
  hope this doesn't cause too many trouble.

==[ v1.0beta2, Fri Jun 17 17:29:03 CEST 2005 ]==
- ::except-hosts wasn't working, reported by Niklas.
- Removed 'y' from all-consonant check, reducing false positives.
- Moved the consonant/vowel/digit checks to a custom function that does
  the counting. This has 2 effects:
  a) much faster than regex, a test on my a1800+ shows that antirandom
     now executes in ~85 microseconds instead of ~700 microseconds for
     the same usernames (this excludes the time spend on the exception
     list btw). Basically this means that the AntiRandom routine alone
     can process roughly ~7000 local users/sec on my amd 1800+ (that
     is: assuming a worst-case 120usec/user [=8333users/sec] and some
     exception list stuff subtracted, and [near] 100% CPU).
     In short, if you got a fast GHz machine, you should be able to
     deal even with the greatest threats (as far as AntiRandom is
     concerned, at least).
  b) we now get proper scores. For example 5 digits get a score of 5,
     6 digits get a score of 6, etc.. In the previous (beta) version
     you might have got a score of 5+6=11 in case of 6 digits.
     Naturally this also means the overall score you get will be lower
     (and more reasonable).
- Added a chapter to the README about performance

==[ v1.0, Tue Jun 21 19:28:05 CEST 2005 ]==
- Made it work on windows (ifdef'ed regexes out, they are not used by
  us anymore anyway).
- Minor README updates
- Released as v1.0

==[ v1.1, Fri Jul 22 03:01:49 CEST 2005 ]==
- Fixed digitscheck not working properly.
- Fixed some checks, they were not using the highest encountered value
  (eg: number of digits) but rather the last found one.
- The above 2 things might have caused quite a few misses, especially
  with numbered bots (eg: ABC21423).