Decimal: A high-performance, arbitrary-precision, floating-point decimal library | Hacker News Reader

Decimal: A high-performance, arbitrary-precision, floating-point decimal library

godoc.org

59 pointsrbanffy9 comments

Interesting! This appears to be implementing Decimal32, Decimal64, and Decimal128 [1].

It's worth remembering that the "first" computer, the ENIAC, used decimal arithmetic. A decimal arithmetic package was one of the first things I wrote for my Z-80 homecomputer as I couldn't imagine anyone working with money using binary floating point! [2] That said, isn't Fintech mostly using scaled integers these days?

[1] https://en.wikipedia.org/wiki/Decimal32_floating-point_forma... [2] My horror was refreshed recently when I compared various mortgage calculators on the web. Not two of them got the exact same amortization schedule :( :(

The documentation says this adheres to the `math/big` API in some respects - can someone speak to what differentiates this from `math/big`?

It deals with floating-point, arbitrary-precision decimal numbers, whereas math/big contains support for arbitrary-precision integers, rationals and binary floating point. Looking at it briefly, it looks like it uses math/big under the hood when the numbers are large, but uses an internal compact form for efficiency when numbers are small ~<= 20 decimal digits.

Yep! That’s what it does. Hopefully I can get it moved over to BCD which will drop basic arithmetic performance, but more than make up for it when faster rounding. :-)

The key word is "decimal", this is a library for decimal floating point arithmetic. `math/big` uses binary floating point. The difference is in which floating point numbers can be represented without requiring an infinite expansion. Lot more info at http://speleotrove.com/decimal/.

(Related topic) Since this might gather eyeballs that would like to see decimal support in JS, I'd be interested to hear more details about use cases people have for this Go package as well as potential use cases / interop in JS.

As a plug for another arbitrary-precision, floating-point decimal library in Go, there's https://github.com/cockroachdb/apd.

Original URL was https://github.com/ericlagergren/decimal, but the doc page has more info.

Odd. I had to spend 5 minutes convincing myself from the site (which uses the same wording) that it was actually a 'decimal floating point' library, as I've never heard such a thing called a 'floating point decimal'.

Reply on news.ycombinator.com