Revision history for Perl extension Math::Int128.

0.21     2015-03-09
      - Remove obsolete compiler checks from C API files (bug report
        and analysis by Matt Sisk)

0.20     2015-02-10
      - Fix syntax error masked by some isspace macro implementations.

0.19     2015-02-08
      - Fix bug #101982: infinite loop in string_to_uint128 (reported
        by Matt Sisk).

0.18     2015-01-05
      - No changes from last release.

0.17_02  2014-12-30
      - Reverted the probe change in the last version. This wasn't necessary
        and it brought back the old failure of clang on FreeBSD i386 claiming
        it supported int128 values when it really didn't.

0.17_01  2014-12-29
      - Trying out a slightly different probe for uint128 support. The
        previous version could have (very rarely) returned false when it
        should have returned true.
      - Made this module compile with the system Perl on OSX by brutally
        stripping the "-arch i386" flag when running the compiler. Your
        results may vary (but it passes its tests).

0.17     2014-12-23
      - No changes from the last release.

0.16_05  2014-12-18
      - More left and right shift fixes. When shifting left by more than 127
        bits, the return value is always 0. When shifting right by more than
        127 bits, the return value is 0 for numbers >= 0 and -1
        otherwise. This is consistent with how Math::Int64 works.

0.16_04  2014-12-15
      - clang on FreeBSD i386 will claim to support 128-bit integers but then
        segfault when compiling this module's XS code. The compiler feature
        checking now uses a more complex test program to ensure that the
        compiler really does support 128-bit integers.

0.16_03  2014-12-13
      - Removed compiler detection in favor of compiler feature checking with
        Config::AutoConf. (Dave Rolsky)

0.16_02  2014-11-05
      - add logic for detecting Apple variant of clang compiler

0.16_01  2014-11-02
      - dump output from compiler version command when running on a
        CPAN tester

0.15     2014-10-30
      - improve clang version detection code (bug report by Olaf
        Alders)
      - relax gcc version detection

0.14     2014-10-29
      - tests were broken on perls compiled with -Dusequadmath (bug
        report by Sisyphus)

0.13     2013-09-07
      - fix unaligned memory access problems happening on
        Windows+Strawerry Perl (bug report by Sisyphus)
      - silent several compiler warnings (bug report and patch by
        d.thomas@its.uq.edu.au)

0.12     2013-07-19
      - improve gcc version detection (bug report by
        d.thomas@its.uq.edu.au)
      - experimental support for clang compiler added

0.11     2013-07-18
      - fix >>= operator (patch by Dave Rolsky)
      - more and better tests (patch by Dave Rolsky)
      - easy compilation of the git version

0.10     2013-07-17
      - add support for exponentiation (bug report by Dave Rolsky)

0.09     2013-07-11
      - operator <<= was broken for unsigned 128bit numbers (bug
        report by Dave Rolsky)

0.08_01  Not Released
      - include stdint.h for 64bit integer definitions (bug report and
        solution by Sisyphus)

0.07     2012-12-10
      - rerelease as stable

0.06_07  2012-11-05
      - gcc version detection was too picky (bug report by Peter John
        Acklam)
      - int128_to_net and uint128_to_net were broken
      - add tests addapted from Math::Int64 module
      - mark internal representation of int128 wrappers as readonly in
        order to disallow their modification from outside the module

0.06_06  2012-07-20
      - update Math::Int64 C API support files

0.06_05  2012-07-19
      - stash caching code was broken on non threaded perls
      - several errors on C API corrected

0.06_04  2012-07-17
      - add support for C API
      - stash caching handling was broken on threaded perls
      - improve die_on_overflow handling

0.06_03  2012-01-11
      - add die_on_overflow feature
      - add support for int64_t integers via Math::Int64 
      
0.05     2011-04-24
      - support for gcc 4.4 and 4.5 contributed by KMX

0.04     2011-03-08
      - delete generated Makefile when aborting Makefile.PL

0.03     2011-03-04
      - abort Makefile.PL unless compiler is GCC 4.6 or better

0.02     2011-02-14
      - add new set of operations that use one argument for output for
        improved performance
      - on string to int128 conversion skip zeros at the left before
        testing for overflow
      - remove OPTIMIZE='-g -O0' from Makefile.PL

0.01     2011-02-07
	- original version; created by h2xs 1.23 with options
		-An Math::Int128