Add decimal support

[MarkSwanson]

I need to add quantum mechanics into the mix. Planck's constant is 6.62607015 × 10−34 joule second.
I've looked at a few arbitrary precision libraries but I'm not sure which one to recommend. I like 'scientific' because it parses e notation. The bigdecimal crate also seems like it would work.

Thanks!

[MarkSwanson]

Fyi f64 supports an accuracy of ~= 1.0 x 10 ^ 16. Folks doing scientific work need around 1.0 x 10 ^ 40.

[iliekturtles]

Does the bigrational (num::BigRational) underlying storage type meet your needs?

If not, I think the next step would be to review arbitrary precision libraries and figure out which to support.

[MarkSwanson]

I tried BigRational but it just doesn't fit - I can't always specify my numbers as a fraction. Of the various arbitrary precision libraries out there the num-decimal crate provides some simple functionality that seems to work. Not sure of what limitations I'll face yet.

[iliekturtles]

A new decimal underlying storage type would be needed. See #287 for a similar effort to add complex numbers. I started a review of options. 3 year old summary: https://www.reddit.com/r/rust/comments/a7frqj/have_anyone_reviewed_any_of_the_decimal_crates/

Most popular options:

https://crates.io/crates/bigdecimal
https://crates.io/crates/rust_decimal
https://crates.io/crates/decimal

Less popular options:

https://crates.io/crates/fraction
https://crates.io/crates/rug
https://crates.io/crates/decimal-rs

[MarkSwanson]

The problem with most of those (I'm not familiar with rug) is that the number type they provide does not support Copy.
Here's why it's bad, and how I worked around it with another unit/quantity crate:

mamrhein/quantities.rs#1 (comment)

Maybe this would be an easier / more ergonomic approach for uom as well.

[iliekturtles]

See #38 for implementing operators for references.

[MarkSwanson]

I took a stab an implemented support for f256 using the quad library
[email protected]:MarkSwanson/uom_num_decimal.git branch=qd_f256

cargo test shows 67 warnings.
It breaks f64 and f32 a tiny bit. I left it like that as it's just a proof of concept.

Also, I'm unable to get even a simple example (examples/qd.rs) to work so it obviously doesn't work.

Any thoughts on why qd.rs gets so many errors?

[iliekturtles]

A number of your changes are to the existing f32 or f64 code. While you might be able to add Quad to the existing storage_types! { types: Float; ... } blocks, you'll probably need new blocks. Once you've done that feel free to open a PR and I can run the full test suite or if you could paste specific commands you're running and the first few errors I can review in more detail.