A C++ port of the C# BigInteger class, with bindings for Python.
The default configuration builds a shared library
mkdir build
cd build
cmake ..
cmake --build .
You can define various options during the configure stage to control the output
Var | Default | Comment |
---|---|---|
BIGINTEGER_BUILD_SHARED_LIB | ON | |
BIGINTEGER_BUILD_STATIC_LIB | OFF | |
BIGINTEGER_BUILD_TESTING | OFF |
Read here.
Test can be ran as follows
cmake -D BUILD_TESTING=ON ..
cmake --build .
./tests/unit_tests
The output should be similar to
...
[ RUN ] properties.RunMinusOneTests
[ OK ] properties.RunMinusOneTests (2 ms)
[----------] 3 tests from properties (7 ms total)
[----------] Global test environment tear-down
[==========] 117 tests from 22 test suites ran. (17481 ms total)
[ PASSED ] 117 tests.
The project can be included in 2 ways
add_subdirectory(BigIntegerCpp)
or if it has been installed via make install
you can use
find_library(bigintegercpp)
It is possible to build both a SHARED
and STATIC
library simultaneously, as such 2 link targets exists to differentiate between them, respectively
BigIntegerCpp::BigIntegerCpp
BigIntegerCpp::BigIntegerCpp-static
To use in your target add
target_include_directories(your_target PUBLIC ${bigintegercpp_INCLUDE_DIRS})
target_link_libraries(your_target PRIVATE BigIntegerCpp::BigIntegerCpp)
Source.cpp
#include <bigintegercpp/BigInteger.h>
-
Why this project?
It is purpose build for use in relationship to the NEO blockchain project. In order to create a compliant port of their virtual machine a need for a compliant BigInteger implementation exists. Any difference, in for example the modulo implementation, can result in VM execution deviation. This is just one of the many problems we've encountered after attemping to wrap Python's native
int
to produce identical behaviour. -
Should I use this project?
If you have to ask this question, then no.
-
How fast is it?
We don't know. The focus has been on conformity to the C# BigInteger class, not on speed. If you want speed you might want to look at https://gmplib.org/
-
Are there any known behavioural deviations from the C# implementation?
The only known deviations are in the string parsing and conversion to string methods. Specifically, the overloads with
IFormatProvider
are not supported. Only base10 parsing is supported. String input may be prepended with+
or-
. Any whitespace is considered the end of the input.Deviations in any other parts are considered bugs. Please report them.