gmperl is a low-level Erlang NIF language binding for the GNU MP Bignum (multiple precision arithmetic library). Tha main goal is to augment Erlang's native bignum support with fast rational and mixed precision floating point numbers.

• Erlang R14B1 or greater
• GMP (tested with 4.3.2 and 5.0.1; mpz_powm_sec requires 5.x)

gmperl's Git repo is available on Github at:

http://github.com/skarab/gmperl

Clone the repo with:

git clone git://github.com/skarab/gmperl.git

# Run bootstrap to generate sources
$ ./bootstrap
# Run EUnit tests
$ ./rebar eunit

1. Fork it, clone your fork, and create a new branch
2. Add code
3. Make sure new code has test that pass
4. Push your branch and send a pull request

gmperl makes use of Erlang's NIF functionality. Note that long-running NIF calls will block other NIF operations (due to the way they are implemented in the Erlang scheduler).

Integer functions are prefixed with mpz_.

1> I = gmperl_nifs:mpz_init().
<<>>
2> ok = gmperl_nifs:mpz_set_ui(I, 10000000).
ok
3> J = gmperl_nifs:mpz_init().
<<>>
4> ok = gmperl_nifs:mpz_set_ui(J, 3).
ok
5> Result = gmperl_nifs:mpz_init().
<<>>
6> ok = gmperl_nifs:mpz_add(Result, I, J).
ok
7> gmperl_nifs:mpz_get_str(Result, 10).
"10000003"

Functions for rational numbers are prefixed with mpq_.

8> Q = gmperl_nifs:mpq_init().
<<>>
9> ok = gmperl_nifs:mpq_set_num(Q, I).
ok
10> ok = gmperl_nifs:mpq_set_den(Q, J).
ok
11> gmperl_nifs:mpq_get_str(Q, 10).
"10000000/3"

Functions for floating-point numbers are prefixed with mpf_.

12> F = gmperl_nifs:mpf_init_set_str("1234.0000001", 10).
<<>>
13> gmperl_nifs:mpf_get_str(F, 10, 100).
{"12340000001",4}
14> gmperl_nifs:mpf_get_prec(F).
64

• Batch allocation for speed
• gmperl module with more Erlang/FP-friendly methods
• Serialization/deserialization of GMP terms

See the GMP documentation at http://gmplib.org.

Copyright (c) 2011 Hunter Morris. See LICENSE for details.