[ANN] A decimals lib like Java BigDecimal

Package Announcements
1
1 Like
2

isaactsang, Are you and gfZeng, one and the same person?

3

(I am assuming this is a package announcement, and changed the category accordingly)

4

Yeah, it’s me, both

5

thanks

6

Could you explain the benefit of this package over using BigFloat or BigInt?

1 Like
7

The real infinite precision support

big"1.0" / big"3.0" 
=> 0.3333333333333333333333333333333333333333333333333333333333333333333333333333348

using Decimals
Decimals.setprecision!(precision=79)
d"1.0" / d"3.0"
=> Decimal(3333333333333333333333333333333333333333333333333333333333333333333333333333333, -79)
8

You will need to add an open-source license, such as MIT, for people to be able to use the package.

4 Likes
9

Thanks for your remind, I have added just now

2 Likes
10

What are the Pros and Cons vs GitHub - JuliaMath/Decimals.jl: Pure Julia decimal arithmetic library.
And
GitHub - JuliaMath/FixedPointNumbers.jl: fixed point types for julia

3 Likes
11

compares with GitHub - JuliaMath/Decimals.jl: Pure Julia decimal arithmetic library.
at here

Well, I don’t know much about GitHub - JuliaMath/FixedPointNumbers.jl: fixed point types for julia

12

Well, the only comparison with Decimals.jl that you give at the referenced link is that your implementation is simpler and the use is more friendly. Now that you made such statement, could you be a bit more specific? In which aspect is your implementation simpler? And in which situations is the use more friendly? Perhaps examples might help.

You may also want to take an inspiration from a comparision displayed on the Decimals.jl page: https://github.com/JuliaMath/Decimals.jl#comparison-with-other-packages

2 Likes
13

The code is simple enough, best docs is code. :slight_smile:

14

yeah,it is, the idea of a string macro is very good. with that said, as i saw a lot of times with number packages, a speed comparison would be very good. the main difference i saw is that the JuliaMath Decimals.jl uses the following struct:

struct Decimal <: AbstractFloat
        s::Integer  # sign can be 0 (+) or 1 (-)
        c::BigInt   # coefficient (significand), must be non-negative
        q::Integer  # exponent
end

your decimal uses:

struct Decimal <: AbstractFloat
    c::BigInt  # coefficient
    q::Integer # exponent
end

a glaring fault is the absence of compatibility with other number types. (for example, 1+d"3.0")

15

1+d"3.0" should be working, you can try again

16

Isaactsang,

I take great offense that your implementation is simpler and the use is more friendly.

Let me give you an example of my PDFPs.jl which is more simpler and more friendly

[ENTER SARCASM MODE]

julia> using PDFPs

julia> s = PDFP_toCommonString
PDFP_toCommonString (generic function with 1 method)

"Now using 4 digit precision via format of PDFP(number,4)";

julia> A=PDFP("7.0",4); B=PDFP(pi,4); C=A+B; s(C)
"10.14"

"Now using the simpler and more friendlier 0 digit precision";

julia> A=PDFP("7.0",0); B=PDFP(pi,0); C=A+B; s(C)
""

julia> show(C)
PDFP(0, 0, [])

Any number can be represented by my 0 digit precision PDFP and the result is also a 0 digit precision PDFP. NOTHING could be simpler or more friendly to use. In fact my 0 digit precision PDFP is the fastest decimal floating point number system in the world because it can give you the result before you even started the calculation!!! It is also 100% foolproof because it is absolutely impossible to get the wrong result.

[LEAVE SARCASM MODE]

17

Well, we should be considering what features we use most.

You can write a simple function to achieve your case, or just use setexponent(d, 0, RoundDown)

19

https://github.com/gfZeng/Decimals.jl/commit/2783c62e2544ca28c1321894e41d6b597f93a7f4