The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.
Make all the data.* files contain only the 1st value.  All others are sorted
correctly automatically (and not stored).  When harvesting, it'll notify of
new/removed elements.

Add option to Recur: ignore holidays
   then flags like NBD ignore holidays (but count weekends)
When calculating holidays, use this option by default to simplify
   holidays.  Then, order of definitions is not important.

Add options:
   onlyiso8601, etc. to parse
   no-delta-secs   (not delta as a plain number)

Can I reduce the number of timezone modules loaded when parsing a date
with an abbreviation or offset by loading one and testing it before
loading a second?

Make sure the following work:
   1*12:0:24:0:0:0*FW1 = Christmas Day (observed)
   December 25 2015    = Christmas Day

Add a new type of recurrence (???):
   *Christmas Day*NWD

########################################################################
# Deprecated variables:

03/01/2016  TZ

########################################################################
# +1 significant release
########################################################################

Check performance if changing
   sub ... {
      return &sub(...)
   }
to
      goto &sub(...)

In Recur.pm, handle encodings in parse

Add delta.parse_LANG tests

Add UseTZ
  = %all      use all timezones
  = %local    use local timezone
  = Z1 Z2 ... use zones Z1, Z2, etc. (one can be %local)
Only affects parsing.

Add ways to get timezone in cygwin

Cache Date::Manip::TZ::zone for ($abbrev,isdst), ($abbrev,$isdst,$offset),
other???

Profile it and look for optimizations.

Date::Manip::Base : get rid of
   _calc_date_time_strings
   _delta_convert

Benchmarks
   Modules
      5.x
      6.00
      6.00 parse_format
      DateCalc
      TimeDate
      ???
   Tests
      10,000 dates (parse)                          time + size
      10,000 dates (parse + 2 adds + 1 unix date)   time + size
      10,000 scripts (parse 1 date + 2 adds + 1 unix date each) time

Rewrite Problems.pod (Date Manip is slow)

Clear out all problems from CPAN

Methods which require a valid object (secs_since_1970_GMT) should exit
instead of trying to perform the operation if the object is invalid.
RT #60662 (Matt Blythe)

########################################################################
# +2 significant release
########################################################################

Everywhere a timezone can be entered, allow:
   zone
   abbrev
   offset
followed by an option:
   std       either STD or DST time, test STD first (default always)
   dst       either STD or DST time, test DST first
   stdonly   only test STD
   dstonly   only test DST

Change Date::Manip::Base so that $date input can be reference or
string.

Clear as much of the backlog of suggestions as possible.

########################################################################
# TO DO
########################################################################

Make sure there is a correspondance between:
  time,localtime,gmtime
  Date_SecsSince1970,Date_SecsSince1970GMT
  UnixDate(...,"%s"),UnixDate(...,"%u")
and document it all.

Support timezones of the format +500.  David Coppit

Make sure that &DateCalc($date1,"") returns an error.  Jim Anderson

Change the Jan1Week1 variable to accept the values "m1-m7" (1st week contains
Jan X) or "d1-d7" (1st week contains the 1st dX day of week ... so d1
means that the 1st week of the year contains the 1st Monday).

Free up the '%u', '%h', and '%X' printf formats. Reserve '%X' for
  extended formats (%Xa, %Xb, ...).

########################################################################
# TO CONSIDER
########################################################################

Multilingual parsing:  Yuri Nikulin
   language=english,french
   regexps then support both languages

Special date formats in language file:
   extra{LABEL} = [ ... ]         extra words of type LABEL
   offset_date =>
      FORMAT => 'OFFSET'
         where FORMAT is similar to parse_format
         %LABEL is a regexp with any of the words
         OFFSET can include %y, %d, %w, %m, %h, %mn, $s in them
   same for offset_time, times, and others
   ex.
      offset_time =>  { "%h o'clock" => "%h:00:00" }
   Support some of the special Russian dates supplied by Yuri Nikulin

Add a method:
   ($date0,$date1) = $date->week_range();
where $date0 and $date1 are the start and end of the week containing
$date.  Ha Quach

Add Date_LocaleInit which calls Date_Init and then sets DateFormat
config varialbe.  Benjamin Low
   Essentially, I use POSIX::strftime to print a known date in the locale
   'native' format ('%x'), and parse the result to determine d/m/y, m/d/y,
   or y/m/d.
   
   Here's what I do for Date::Parse, perhaps for your module you could just
   substitute a default value for DateFormat:
   
   sub _dmorder
   # determine the "natural" day/month order for the current locale
   # - returns a sub which will expect two arguments (month, day) and
   #   return the arguments swapped as appropriate
   {  
	# %x - preferred (year, month, day) representation
	# - some examples: 1999-12-31, 31/12/99, 30.12.1999, 12/31/99
	my @d = (POSIX::strftime('%x', 0, 0, 0, 31, 12-1, 99) =~ 
		/(\d+)\D+(\d+)\D+(\d+)/);

	# check we got one each of "31", "12", and "[19]99" back
	$@ = "couldn't determine day,month order (got [@d])";
	warn("$@\n"), return sub { @_ } unless @d == 3;
	my %d;	$d{$1} = $d{$2} = $d{$3} = 1;
	warn("$@\n"), return sub { @_ } 
		unless ($d{31} and $d{12} and ($d{99} or $d{1999}));

	if ($1 == 31) { $@ = undef; return sub { ($_[1], $_[0]) } };	# d/m/y
	if ($2 == 31) { $@ = undef; return sub { ($_[0], $_[1]) } };	# m/d/y
	if ($3 == 31) { $@ = undef; return sub { ($_[0], $_[1]) } };	# y/m/d

	return sub { @_ };	# undetermined, use default
   }
   
   *dmorder = _dmorder();
   
   # and then later in Parse::Date, after month/day regexps (\d+/\d+)...
     - ($month, $day) = ($1, $2);    becomes...
     + ($month, $day) = dmorder($1, $2);


Make DateFormat variable handle y/m/d y/d/m m/d/y and d/m/y formats
in addition to m/d vs. d/m .  Also, make "%D" and "%x" UnixDate formats
use this variable.  Benjamin Low

Make the following work for ParseDate Adrian Conte:
  1 epoch
  epoch 1
  -1 epoch
  epoch -1

Make work weeks able to start and stop on arbitrary days (even across
weekends).  Mohammed Saggaf

Switch to Math::BigInt instead of using "no integer".  Vishal Bhatia

Use autoloader.  Ted Ashton

Better support for fractional seconds.  RT 61535

########################################################################
# GRANULARITY
########################################################################

$flag=&Date_GranularityTest($date,$base,$granularity [,$flags] [$width])
   $date and $base are dates
   $granularity and $width are deltas
   $flags is a list of flags

   To test if a day is one of every other Friday (starting at Friday
   Feb 7, 1997), go:
      $base=&ParseDate("Friday Feb 7 1997");
      $date=&ParseDate("...");
      $granularity=&ParseDateDelta("+ 2 weeks");
      $flag=&Date_Granularity($date,$base,$granularity,"exact");
   If $flag is 1, the $date is a 2nd Friday from Feb 7.

   The most important field in $granularity is the last non-zero element.
   In the above example, 2 weeks returns the delta 0:0:14:0:0:0 so the
   last non-zero element is days with a value of 14.

   If $flags is empty, $date is checked to see if it occurs some multiple
   of 14 days before or after $base.  In this case, hourse, minutes, and
   seconds are completely ignored.

   If $flags contains the words "before" or "after", $date must come
   before or after $base.

   If $flags contains any other options, or if $width is passed in, the
   test is treated in an approximate way.  A flag of "approx" forces this
   behavior.

   If $width is not passed in in an approximate comparison, it defaults
   to 1 in the last non-zero element.  Here, the default width is 1 day.
   If the flag "half" is used, the width (default or passed in) is
   halved.

   For example if $width is 1 day, add a multiple of $granularity to
   $base to get as close to $date as possible.  If $date is within plus
   or minus 1 day of this new base, the test is successful.  A flag of
   "plus" or "minus" means that $date must be with plus 1 day or within
   minus one day of this new base.  Flags of "before" or "after" work
   as well.

@list=&Date_GranularityList($date,$N,$granularity)
   Returns a list of $N dates AFTER $date which are created by adding
   $granularity to $date $N times.  If $N<0, it returns $N dates BEFORE
   $date (the list is in chronological order).

   Make it work in business mode as well which will return only working
   days.  Example, every other friday and it can be told that if friday
   falls on a holiday to return either thursday or the following monday
   or leave it out.