# Changelog
## 0.4.4
- Bump MSRV from 1.61 to 1.68.
## 0.4.3
- Mark `FBig::from_str_native` as deprecated.
- Implement `TryFrom<Repr>` and `TryFrom<FBig>` for primitive integers.
- Implement `TryFrom<Repr<2>>` and `TryFrom<FBig<_, 2>>` for primitive floats.
- Implement `From<UBig>` and `From<IBig>` for `Repr`.
- Implement `core::fmt::{Binary, Oct, LowerExp, UpperExp, LowerHex, UpperHex}` for `Repr`, `FBig` (some are limited to certain bases).
## 0.4.2
- Add `Repr::from_static_words` to support the `static_fbig!` and `static_dbig!` macros.
- Add `FBig::from_repr_const` to support create an `FBig` instance from repr in const context.
- Add conversion from `f32`/`f64` to `Repr<2>`.
- Implement `NumOrd` between `FBig` and primitive integers / floats.
- Implement `AbsOrd` between `FBig` and `UBig`/`IBig`.
- Now the `Debug` output of `FBig` values will not contains the rounding mode information (when alternative flag is not set).
## 0.4.1
- Fix the termination criteria for `ln` and `exp` series ([#44](https://github.com/cmpute/dashu/issues/44)).
- Fix `powf` panicking when base is 0.
## 0.4.0
### Add
- Implement `num-order::NumOrd` between `FBig` and `UBig`/`IBig` and between `FBig` with different bases.
- Implement `num-order::NumHash` for `FBig` and `Repr`.
- Add `ErrorBounds` trait that calculate the rounding range for a floating point number.
### Change
- Now feature `num-traits` and `rand` are not enabled by default, feature `num-order` is enabled instead.
- The type of `Repr::BASE` is changed from `IBig` to `UBig`
- `UBig::square` and `IBig::square` are renamed to `sqr`.
- The implementation of square root is now implemented by the `dashu_base::SquareRoot` trait instead of a standalone method of `FBig`.
- The rounding behaviors of `FBig::to_decimal` and `FBig::to_binary` are changed for better ergonomics.
- The rounding behaviors of `FBig::to_f32` and `FBig::to_f64` now follow the mode specified by the type argument.
## 0.3.2
- The default precision for float numbers from `from_parts`/`From<UBig>`/`From<IBig>` are now based on the actual digits on the integers, rather than the digits after simplification. (#28)
## 0.3.1
- Implement `num_traits::{Zero, One, FromPrimitive, ToPrimitive, Num, Signed, Euclid, Pow}` for `FBig` (#19)
- Implement `rand::distributions::uniform::UniformSampler` for `FBig` through `crate::rand::UniformFBig`
- Implement `rand::distributions::{Open01, OpenClosed01, Standard}` for `FBig`
- Implement `dashu_base::Inverse` for `FBig`
- Implement `rand::distributions::uniform::SampleUniform` for `FBig`.
- Implement `serde::{Serialize, Deserialize}` for `FBig` and `Repr`
- Implement `Rem` trait for `FBig`
- Add support of random floating point numbers generation through `crate::rand::Uniform01` and `crate::rand::UniformFBig`.
- Add support for serialization from/to PostgreSQL arguments through `diesel::{deserialize::FromSql, serialize::ToSql}` and `postgres_types::{FromSql, ToSql}`.
- Add `from_str_native()` for `Repr`
- Add `to_f32()`, `to_f64()` for `Repr`, and these two methods supports all bases for both `Repr` and `FBig`.
- Add `to_int()` for `Repr`, which is equivalent to `FBig::trunc()`
- Add `TryFrom<FBig>` for `UBig` and `IBig`
- Add `round()` for `FBig`
- Add `rand_v08` and `num-traits_v02` feature flags to prevent breaking changes due to dependency updates in future
- Re-export operation traits through the `ops` module.
## 0.3.0
### Add
- Conversion from FBig to `f32`/`f64` support subnormal values now.
- Add a `split_at_point()` function to `FBig`
## 0.2.1
- Implement `core::iter::{Sum, Product}` for `FBig`
- Implement `powf`, `sqrt` for `FBig`
## 0.2.0 (Initial release)
- Support basic arithmetic operations (`add`/`sub`/`mul`/`div`/`exp`/`ln`) and base conversion.
# Todo
## Roadmap to next version
- Support generating base math constants (E, Pi, SQRT2, etc.)
- Support operations with inf
- Create operations benchmark
- Benchmark against crates: rug, twofloat, num-bigfloat, rust_decimal, bigdecimal, scientific
- Implement more formatting traits
- Other math functions: sin/cos/tan/etc.