1.6611: Wed Dec 17 2014
   - SREZIC@cpan.org suggested LC_ALL=C for all tests
   - using Number::Format vars instead of LC_ALL to fix tests
   - yeah, that's not going to work... using setlocale()

1.6608: Fri Jan 10 2014
   - bug in LSF

1.6607: Mon Jan 23 2012
   - forgot to update the copyright to 2012

1.6605: Sat Jan 21 2012
   - change env var DEBUG to DEBUG_STATS_B (cpan698301@mstier.de)

1.6603: Mon Jan 24 2011
   - fail to load Test::More gracefully, but don't require it
     even for build.

1.6602: Thu Sep 30 2010
   - changes to make this work again in 5.6.x
     (hint provided by Jerome Provensal)

1.6601: Sat Sep 12 2009
   - The debian build devs set TEST_AUTHOR, then created a bug
     report concerning the results.  First things first, make it
     so setting TEST_AUTHOR doesn't work unless you set a
     specific code -- are you the author? No?  Then don't
     complain about the suggestions from Perl-Critic
   - fixed the various things the latest Perl-Critic complains
     about -- nothing really, and a bug in the critic.

1.6600: Sun Jun 28 2009
   - new test: create each object in its empty state, make sure
     its really empty.
   - added an importer interface for the %ENV vars
   - removing the ENV vars (except for pre-compile init)
   - finished updating the pods for all the 1.66 changes

1.6600: Sat Jun 27 2009
   - worked on computed vectors a little.  decided it should be
     possible to insert into them, instead inserting into the
   - finished a lot more 1.66 docs
   - since two vectors remain separate objects under
     $v1->set_vector($v2), they shouldn't be linked together from
     that -- set_vector now copies instead of linking.
     computed() vectors can serve that role instead.
   - made sure all the object accessors return the object
   - made set_vector() take nonscalars like all the export
   - made the export constructors use the new set_vector() via
   - made all the objects true in bool context

1.6600: Wed Jun 24 2009
   - _OVBed: median, StdDev, Variance
   - fixed bugs resulting from the above
   - made quite a lot of the methods of Vectors private
   - trying to make the pods more consistent and informative
   - renamed size() as query_size() to be really consistent --
     but put in an undocumented alias named size() ... maybe
     deprecate later.
   - renamed ginsert() append(), but left ginsert() operational,
     with no real chance of deprecation later

1.6600: Tue Jun 23 2009
   - installed a lotta DRY on the two vector classes, will
     continue this tonight and on lunch break most likely.
   - worked on the docs, they need a good deal of work
   - _OVBed Mean
   - noticed a half-way implemented set_vector fill-disabler, why
     isn't it finished?
     - added a query_filled to check to see if a vector is full
     - made sure the _OVB and _TVB base classes know about the
       half implemented fill disabler
     - use query_filled in various places, to make it automagic
       like everything else
     - I gave up on the second argument to set_size... it needs
       to be kinda system wide to work right.  Another $ENV{VAR}.
   - _OVBed Mode

1.6600: Mon Jun 22 2009
   - started added best practices stuff to the dist

1.6500: Fri May 29 2009
   - I get asked, "does S::B handle missing values?" quite a lot.
   - YES IT DOES! (now)

1.6007: Sat May  2 2009
   - fixed copyright information on request
   - fixed various docbugs
   - finished a thought about outliers

1.6005: Thu Mar 26 12:22:23 EDT 2009
   - META.yml stuff

1.6004: Sun Mar  1 07:15:37 EST 2009
   - doc bug found by Raj Chandran

1.6003: Sun Feb  8 17:21:39 EST 2009
   - perl 5.8.0 can't deal with this program, requiring 5.8.1:

     my $o = bless {v1=>1, v2=>2}, "class";
     print "K:$_; V:$o->{$_}\n" for keys %$o;
     print "K:$_; V:$o->{$_}\n" for qw(v1 v2 v3);

1.6002: Mon Jan 19 07:16:55 EST 2009
   - I'm skipping the t/60_ tests without recent versions of
     Math::BigFloat.  No idea what causes the problems, but since
     the module isn't required, it doesn't even tell me the
     version of the module.
   - Perl's locale support is embarrassingly buggy.  Under
     certain conditions, Number::Format will die because perl's
     POSIX getlocale stuff returns some (but not all) of certain
     locales.  This stuff is reproducible on almost any
     machine, so I didn't bother to report it.  WRW installed a
     fix in NF 1.61(a) that sets the thousands_sep to "" when
     it's the same as the decimal_point.  Fixed (sorta).

1.6001: Thu Nov  6 07:06:24 EST 2008
   - noted the population vs sample issue (again)

1.6: Sun Oct 19 07:24:19 EDT 2008
   - rather than fixing more of the tests using
     abs(blah-blarg)<tolerance, I've moved to overload '==' to do
     it for me iff $ENV{TOLER} is set to a tolerance.
   - I rather trivially allow Math::BigFloat inserts, but this
     isn't documented or tested well.

1.531: Thu Oct 16 16:26:12 EDT 2008
   - A really minor bugfix thanks to Michiel van Loon
   - I'm also requiring the latest version of Number::Format to
     see if that mysterious separator problem disappears for
     the rare tester that has a problem with it.

1.53: Fri Aug 22 17:34:02 EDT 2008
   - did some cleanup on the recalc code
   - still showing a bug in the new t/08_filter tests... 
   - yeah, bad bug in the median code, wow

1.50: Fri Aug 22 06:56:17 EDT 2008
   - adding a computed vector class that will act just like a
     vector, but will be recomputable.
   - computed vectors and a test have been added...
   - added a filtered-outlier test
   - documented the new computed vectors

1.00: Thu Aug  7 07:24:36 EDT 2008
   - finished up the docs, they still need to be proofread

1.00: Fri Aug 1 - Wed Aug 6 2008
   - bringing this distro up to date, apparently people actually
     use it.
   - tons of doc fixes
   - adding an overloader class to deal with "" and 0+
   - pre-reqed in Number::Format for ""
   - forget returning undef cardinality, vector sizes are 0 when
     they don't have a size set.
   - added vector tags and the ability to add "computers" linked
     to other vectors
   - renamed CoVariance to Covariance
   - I modernized and updated the crap out of the vector class
   - I tested most of that new stuff also
   - I added the "computer" storage devices to the Vector class
   - I started the new smarter linking
   - fixed an unimportant memory leak with Scalar::Util::weaken()
   - re-wrote the docs from scratch

0.42:  Wed Jan 25 17:22:11 EST 2006
   - jettero patched in Orien's things

0.41.3:  Thu Mar 25 12:48:56 EST 2004
   - jettero took out the stoopid mod::sig, everyone hated
     it anyway

0.41.1 and 0.41.2: Tue Mar 23 07:59:46 EST 2004
   - Copyright clarifications.  ... GPL with a twist.

0.41:  Fri Jan 30 08:54:37 EST 2004
   - jettero found an amazingly STUPID typo in the Basic.pm
     Statisics anyone?

0.4:  Thu Jan 29 11:20:51 EST 2004
   - jettero added the UNBIAS flag for those that wish to
     use (N-1) in their variance. Perhaps I should switch
     the default. We will see.

0.39:  Fri Dec 12 07:18:18 EST 2003
   - jettero changed the NAME field to match CPAN

0.38:  Tue Dec  9 08:35:59 EST 2003
   - jettero discovered many bugs in the insert/ginsert of
     the new new()s

0.37:  Mon Dec  8 20:58:08 EST 2003
   - jettero cleaned up the new() functions so they don't
     require arguments for no reason

0.36:  Sat Dec  6 11:32:17 EST 2003
   - jettero added (linear) LeastSquareFit

0.35:  Tue Dec  2 17:04:58 EST 2003
   - jettero was working on the documentation in Basic.pm
     when he realized the need for ginsert and set_vector
     around the world... finished implementing those.

0.34:  Tue Dec  2 14:45:11 EST 2003
   - jettero tought the ::Covariance module to take ::Mean
     objects as optional arguments to keep the ::Correlation
     object from uselessly recalculating the mean as
     mentioned in the entry below in 0.33.

0.33:  Tue Dec  2 14:00:48 EST 2003
   - jettero did an amazing amount of work on the objected
     oriented structrues, the tests in t/ and added the
     queuing functions to the packages. He also added the
     Vector object to handle some things. Note that the
     author is aware that the Mean get's uselessly
     recalculated for some the high level objects...
     Solution forthcomming...

0.25:  Mon Dec  1 12:13:44 EST 2003
   - jettero installed a correlation module and upped the
     version a great deal.

0.02:  Mon Dec  1 00:28:50 EST 2003
   - jettero made some base packages, namely: StdDev,
     Variance, Covariance, and Mean.

0.01:  Sun Nov 30 11:55:03 EST 2003
   - jettero made a skeleton package