```
Revision history for Perl extension Math::BigInt::FastCalc.
2014-03-04 0.31 pjacklam
* See http://perl5.git.perl.org/perl.git/history/HEAD:/dist/Math-BigInt-FastCalc
for a list of recent changes.
2011-09-04 0.30 pjacklam
* Math::BigInt::FastCalc now requires Math::BigInt v1.997 (Peter John Acklam).
* Include "^MYMETA\.(yml|json)\z" in MANIFEST.SKIP. Whereas META.* are
generated by the distribution author at packaging time, MYMETA.* are
generated by the end user at configure time after any dynamic
dependencies are known. (Peter John Acklam)
* Updated common test files from the Math::BigInt distribution. (Peter
John Acklam)
2011-07-09 0.29 rafl (7560 tests)
* Synchronise with blead perl.
+ Reduce object code size significantly through PERL_NO_GET_CONTEXT.
2011-02-26 0.28 pjacklam (7560 tests)
* Sync codebases (CPAN vs. blead): include t/mbi_rand.t from blead,
fix typos in FastCalc.xs and t/bigintfc.t (Peter John Acklam).
* Remove redundant Math::BigInt::FastCalc version info from diagnostic
output of 01load.t (Peter John Acklam).
2011-02-26 0.27 pjacklam (7560 tests)
* Add Math::BigInt::Calc version information to diagnostics output of
01load.t. This is useful for debugging, since Math::BigInt::FastCalc
relies heavily on Math::BigInt::Calc (Peter John Acklam).
* Change bigintpm.inc to reflect recent changes in the Math::BigInt
distribution (Peter John Acklam).
2011-02-08 0.26 pjacklam (7560 tests)
* Add Perl v5.6 compatibility code to FastCalc.xs (RT #63859)
(Peter John Acklam).
* Add 00sig.t for testing SIGNATURE, and 01load.t for basic module loading
and giving diagnostics useful for bug tracking (Peter John Acklam).
* Rename pod.t to 02pod.t and pod_cov.t to 03podcov.t and use more generic
code (Peter John Acklam).
2011-02-06 0.251 pjacklam (7558 tests)
* Add requirement Math::BigInt version 1.991 (Peter John Acklam).
2011-02-05 0.25 pjacklam (7558 tests)
* Remove _new() from XS code and use _new() in Math::BigInt::Calc
instead. The XS _new() doesn't handle integers that are too large to
be represented exactly as floating point numbers, but small enough
to be represented as 64 bit integers. The XS _new() also returns
nan, not inf, when it overflows. This closes RT #63335 and RT #49569
(Peter John Acklam).
* Copied test files from latest Math::BigInt (v1.991): bigfltpm.inc,
bigfltpm.t, bigintpm.inc, bigintpm.t, mbi_rand.t, mbimbf.inc,
mbimbf.t (Peter John Acklam).
2010-11-07 0.24 rafl (6938 tests)
* Reduce repetition (and object size) in FastCalc.xs by using ALIASes
(Nicholas Clark).
* inline the macros CONSTANT_OBJ and RETURN_MORTAL_BOOL. (Nicholas Clark).
* Avoid creating two entries instead of one on the mortals stack in _new() and
_zero() (Nicholas Clark).
* Convert from DynaLoader to XSLoader (Nicholas Clark).
* Adapt to the changed order of the return value Math::BigInt::Calc::_base_len
in Math::BigInt 1.97 (Nicholas Clark).
* Initialise BASE and BASE_LEN as part of bootstrap, avoiding special one-shot
setter routines (Nicholas Clark).
* Remove the $BASE or $BASE_LEN globals. They were never documented or
exported, and are now unnecessary (Nicholas Clark).
2010-09-14 0.22 rafl (6938 tests)
* Modernize the test suite to use Test::More instead of Test.
2010-09-03 0.21 rafl (6892 tests)
* Remove some unneeded test code which only broke things when being ran in the
perl core.
2010-09-03 0.20 rafl (6892 tests)
* Avoid certain kinds of segfaults as well as wrong bnok() behaviour by
depending on Math::BigInt 0.90.
2008-04-15 0.19 Tels (6892 tests)
* forcefully regenerate META.yml (sigh@Module::Install)
2008-04-15 0.18 Tels (6892 tests)
* fix compilation under v5.6.2 (sigh)
2008-04-13 0.17 Tels (6892 tests)
* forgot SvIOK() to SvUOK() (bug #29720, Thanx Alexey Tourbin!)
2007-09-16 0.16 Tels (6892 tests)
* require Math::BigInt 1.88
* disable prototypes
* fix SvIV() vs. SvUV() in _new() (fix bug #29720)
* it works back to v5.006 (thanx zefram!)
2007-06-30 0.15 Tels (6892 tests)
* require Math::BigInt 1.87
* add _nok()
* bundle and use Module::Install
* put _new() into the XS
2007-04-17 0.14 Tels (6438 tests)
* require Math::BigInt 1.83
* support api_version() 2 by adding _1ex() and _alen()
* add a few tests
2007-04-09 0.13 Tels (6280 tests)
* require Math::BigInt 1.82
* revert the accidentily change to "NV elems" in _len
(basically, apply the fix that was already in blead to the CPAN release)
* plug all the leaks in _acmp(), _is_ten() etc. (Thanx Hugo!)
2007-03-03 0.12 Tels (6280 tests)
* fix bug #25033 It leaks (Thanx M. Schwern!)
* require Perl 5.6.2, anything older is unsupported and untested
2007-02-02 0.11 Tels (5953 tests)
* require Math::BigInt 1.79
* add support for _as_oct() and _from_oct()
* add _length() to the method list
Older changelog:
0.01 Wed Jun 5 13:03:55 2002
- original version; created by h2xs 1.21 with options
-n Math::BigInt::FastCalc -A
2002-08-09 0.02 Tels
- first working version of XS code of _is_even(), _is_odd()
2002-08-12 0.03 Tels
- Took Calc.pm and converted it to FastCalc.pm
- first two functions, _is_odd() and _is_even() in XS
- testsuite from MBI v1.61
- extended bigintc.t (also renamed to bigintfc.t)
2002-08-13 0.04 Tels first release
- added is_zero(), is_one(), _zero(), _one(), _two() and _acmp()
- more tests in bigintfc.t
Some general observations:
* is_*() methods are quite a bit faster than in Calc, but due to the
overhead in Math::BigInt real programs will get only a bit faster
(around 2-5%).
* Comparisation is *much* faster, and even via Math::BigInt improves the
performance by about factor 9 for large numbers. For smaller numbers (less
than 8000 digits) it is still about factor 2.5 to 8. This is only valid if
the numbers have both the same length. If the length is different, then
FastCalc is still about a factor of 1.2 to 1.7 faster than Calc.
2002-08-13 0.05 Tels (5224 tests)
* the note that "_len() and _acmp() are in XS" was missing
* cleaned distribution of unwanted files
2002-08-25 0.06 Tels (5355 tests)
* working _modinv() and fix for _rsft() taken over from Calc v0.32
* added tests for _rsft() and _modinv()
* require Math::BigInt v1.62
* _num(), _dec() and _inc() are now in XS
$MBI->numify() is about 50% faster
$MBI->binc()/$MBI->bdec() are about 5-8% faster for numbers that don't
overflow (much), and about 14% for ('9' x 100)->binc()
* __strip_zeros() is no in XS, making it roughly 3 times as fast
* added _copy(): more than twice as fast for large (> 1000 digits) numbers:
Math::BigInt::Calc v0.32
1: 3s ( 3.23 usr + 0.00 sys = 3.23 CPU) @ 38128/s (n=123154)
100: 3s ( 3.19 usr + 0.00 sys = 3.19 CPU) @ 27397/s (n=87397)
1000: 4s ( 3.29 usr + 0.00 sys = 3.29 CPU) @ 8042/s (n=26460)
10000: 4s ( 3.19 usr + 0.01 sys = 3.20 CPU) @ 984/s (n=3149)
Math::BigInt::FastCalc v0.06
1: 3s ( 3.12 usr + 0.01 sys = 3.13 CPU) @ 39346/s (n=123155)
100: 3s ( 3.10 usr + 0.00 sys = 3.10 CPU) @ 34138/s (n=105830)
1000: 3s ( 3.09 usr + 0.00 sys = 3.09 CPU) @ 15807/s (n=48844)
10000: 3s ( 3.20 usr + 0.00 sys = 3.20 CPU) @ 2362/s (n=7559)
2004-08-13 0.07 Tels (5953 tests)
* require BigInt v1.71, adapted to new API style by taking over Calc v0.41
+ added _is_two(), _is_ten(), and _ten()
* cleaned XS code, got rid of "unused var" warnings (thanx perl-xs people!)
2005-01-01 0.08 Tels (5953 tests)
* take tests from BigInt v1.74 and require it
* remove needless Exporter in FastCalc
* Makefile.PL: add writing of own meta file with correct licence
* take over changes from Calc 0.44
2005-03-20 0.09 Tels (5953 tests)
* take over changes from Calc 0.45 (fix bug with div shortcut)
* require Math::BigInt v1.75 and use its tests
2005-03-30 0.10 Tels (5953 tests)
* load Calc.pm and use its code to provide the routines that FastCalc
does not yet have in XS (cut down Copy&Paste code considerably
* require Math::BigInt v1.76 to properly load Calc.pm
```