- Allow for compiler bug (gcc-4.4.3) in Strawberry Perl versions 5.12.x.
   See https://rt.cpan.org/Ticket/Display.html?id=141090

 - Fix typo in Complex_C.xs that prevents this module from building on case-sensitive
   filesystems. ("C" ne "c".)

 - Fix https://github.com/sisyphus/math-decimal64/pull/1 (also aplies to Math::Complex_C)
   Thanks to @hiratara
 - Fix overload_pow() for the case that the third argument is true
 - Replace SvIV/SvUV/SvNV with SvIVX/SvUVx/SvNVX where appropriate.

 - Add nnumflag(), set_nnum() and clear_nnum() functions. Also start using the perl
   API function looks_like_number() to detect non-numeric strings.
 - Change default ouptput precision from 15 to 17 decimal digits.

 - Try to detect if references to "cpow" will not be resolved - and abort the build
   if this problem is found to exist. (This enables us to avoid some FAIL reports
   from cpan-testers.)
 - Rewrite tests 4 and 5 in t/overload_string.t so that they don't register a fail
   if the libc bug that causes 'nan nan' to be returned (instead of the correct
   'inf inf') is present.

 - Remove is_neg_zero() and get_neg_zero(). (Sick of the bugs with this.)
 - Rename Math::Complex_C::Long to Math::Complex_C::L and move to a separate distro.
 - Add metadata to Makefile.PL
 - Remove negative NaN support. Makes no sense to provide a sign to something that
   is not a number ... next thing they'll be wanting is red nans and green nans ....

 - Define __USE_MINGW_ANSI_STDIO in Complex_C.xs for the MinGW compiler so that
   signed zero errors are avoided under Windows XP (and, presumably, earlier).
   This symbol was already being defined in Long.xs.

 - Fix bug in t/arith.t and t/arithl.t
 - Use float.h's DBL_DIG and LDBL_DIG to set the initial value of precision
   for "printf".
 - Add _DBL_DIG() and _LDBL_DIG() XSubs so that we can see the value of float.h's

 - Add Math::LongDouble conversions(LD2cl, imag_cl2LD and real_cl2LD).
 - Alter the way that strEQ gets hold of HvNAME(SvSTASH(SvRV(sv))) - so that -Wall
    doesn't emit "null argument" warnings. (Thanks Daniel Kahn Gillmor.)
 - Add precision setting/getting functions - d_set_prec, d_get_prec, long_set_prec
   and long_get_prec. This enables control over precision presented by the various
   stringification sub routines (d_to_str, d_to_strp, ld_to_str, ld_to_strp and
 - Add to_string conversions - d_to_str, d_to_strp, ld_to_str and ld_to_strp.
 - Change the print overloading (sub _overload_string) to use the new to_string
   conversion functions. This standardises the printing of double and long double
   objects, and handles the printing in the same way as Math::LongDouble does.

 - Fix is_inf() and is_infl() in Complex_C.xs and Long.xs respectively.
  (They were reporting a NaN to be an Inf.)

 - Remove 'not' from overloads. Apparently achieved nothing anyway, and 5.16
   warns that it's invalid.
 - Devel::Peek::Dump() a couple of variables in t/arith.t, to try and see
   why tests 10-12 fail - ie iff they fail && if Devel::Peek is available.
   (Only reported failure I know of is

 - Add is_neg_zero() function. (Perl will often present '-0' as '0'.)
 - Add aab_neg_zero.t to check how the compiler deals with -0, and whether
   is_neg_zero() functions correctly.

 - Fix assign_cl - internally it was assigning a double instead of a long double.
 - Skip (rather than fail) tests that are known to be failing because
   of compiler bugs.
 - Change _overload_equiv to cater for comparison with real numbers, as
   well as complex numbers.
 - Move the files in Long/t to top-level t folder, and delete Long/t.

 - Remove Long/t/aab_basic.t. (Served no additional purpose.)
 - Add Long/Long.xs to MANIFEST.

 - First release.