This is a Go binding around C decNumber package, for calculation with base-10 floating point numbers. Decimal data type is important for financial calculations.
The C decNumber package can be found at: http://speleotrove.com/decimal/
The decNumber package used is "International Components for Unicode (ICU)" version.
Documentation of original C decNumber package
More specifically, you should read these topics:
- Context: http://speleotrove.com/decimal/dncont.html
- decQuad: http://speleotrove.com/decimal/dnfloat.html
- decQuad example: http://speleotrove.com/decimal/dnusers.html#example7
The original C decNumber package contains two kinds of data type:
- decNumber, which contains arbitrary-precision numbers. Storage will grow as needed.
- decQuad, decDouble, decSingle, which are fixed-size data types. They are faster than decNumber.
This Go package
The Quad type contains a 128 bits decimal floating point value, and a 16 bits status field.
This Go package only uses the decQuad data type (128 bits), which can store numbers with 34 significant digits. It is very much like the float64, except that its precision is better (float64 has a precision of only 15 digits), and it works in base-10 instead of base-2.
I have only written the following files:
The other .c and .h files in the directory come from the original C decNumber package.
The code of this Go wrapper is quite easy to read, and the pattern for calling C function is always the same. Parameters are always passed from Go to C as value, and in the other direction too. Strings passed from C to Go are are also passed by value, as array in struct. Strings passed from Go to C using C.CString(s).
go get github.com/rin01/decnum
The test file mydecquad_test.go is very instructive.
Run the test:
All files in this package are under ICU License (see ICU-license.html).