reflow some docs

This commit is contained in:
Trevor Spiteri 2021-04-20 12:57:09 +02:00
parent 2ca975ee46
commit 353ece484e
3 changed files with 403 additions and 475 deletions

246
README.md
View File

@ -30,10 +30,9 @@ range 0.5 ≤ <i>x</i> < 0.5 for signed numbers like
<code>[FixedI32]\<[U32]></code>, and in the range 0  <i>x</i> < 1 for unsigned
numbers like <code>[FixedU32]\<[U32]></code>.
In version 1 the [*typenum* crate] is used for the fractional bit
count `Frac`; the plan is to to have a major version 2 with [const
generics] instead when the Rust compiler support for them is powerful
enough.
In version 1 the [*typenum* crate] is used for the fractional bit count `Frac`;
the plan is to to have a major version 2 with [const generics] instead when the
Rust compiler support for them is powerful enough.
The main features are
@ -46,46 +45,42 @@ The main features are
This crate does *not* provide general analytic functions.
* No algebraic functions are provided, for example no `sqrt` or
`pow`.
* No trigonometric functions are provided, for example no `sin` or
`cos`.
* No other transcendental functions are provided, for example no
`log` or `exp`.
* No algebraic functions are provided, for example no `sqrt` or `pow`.
* No trigonometric functions are provided, for example no `sin` or `cos`.
* No other transcendental functions are provided, for example no `log` or
`exp`.
These functions are not provided because different implementations can
have different trade-offs, for example trading some correctness for
speed. Implementations can be provided in other crates.
These functions are not provided because different implementations can have
different trade-offs, for example trading some correctness for speed.
Implementations can be provided in other crates.
* The [*fixed-sqrt* crate] provides the square root operation.
* The [*cordic* crate] provides various functions implemented using
the [CORDIC] algorithm.
* The [*cordic* crate] provides various functions implemented using the
[CORDIC] algorithm.
The conversions supported cover the following cases.
* Infallible lossless conversions between fixed-point numbers and
numeric primitives are provided using [`From`] and [`Into`]. These
never fail (infallible) and do not lose any bits (lossless).
* Infallible lossy conversions between fixed-point numbers and
numeric primitives are provided using the [`LossyFrom`] and
[`LossyInto`] traits. The source can have more fractional bits
than the destination.
* Checked lossless conversions between fixed-point numbers and
numeric primitives are provided using the [`LosslessTryFrom`] and
[`LosslessTryInto`] traits. The source cannot have more fractional
bits than the destination.
* Checked conversions between fixed-point numbers and numeric
primitives are provided using the [`FromFixed`] and [`ToFixed`]
traits, or using the [`from_num`] and [`to_num`] methods and
[their checked versions][`checked_from_num`].
* Fixed-point numbers can be parsed from decimal strings using
[`FromStr`], and from binary, octal and hexadecimal strings using
the [`from_str_binary`], [`from_str_octal`] and [`from_str_hex`]
methods. The result is rounded to the nearest, with ties rounded
to even.
* Infallible lossless conversions between fixed-point numbers and numeric
primitives are provided using [`From`] and [`Into`]. These never fail
(infallible) and do not lose any bits (lossless).
* Infallible lossy conversions between fixed-point numbers and numeric
primitives are provided using the [`LossyFrom`] and [`LossyInto`] traits.
The source can have more fractional bits than the destination.
* Checked lossless conversions between fixed-point numbers and numeric
primitives are provided using the [`LosslessTryFrom`] and
[`LosslessTryInto`] traits. The source cannot have more fractional bits than
the destination.
* Checked conversions between fixed-point numbers and numeric primitives are
provided using the [`FromFixed`] and [`ToFixed`] traits, or using the
[`from_num`] and [`to_num`] methods and [their checked
versions][`checked_from_num`].
* Fixed-point numbers can be parsed from decimal strings using [`FromStr`],
and from binary, octal and hexadecimal strings using the
[`from_str_binary`], [`from_str_octal`] and [`from_str_hex`] methods. The
result is rounded to the nearest, with ties rounded to even.
* Fixed-point numbers can be converted to strings using [`Display`],
[`Binary`], [`Octal`], [`LowerHex`] and [`UpperHex`]. The output
is rounded to the nearest, with ties rounded to even.
[`Binary`], [`Octal`], [`LowerHex`] and [`UpperHex`]. The output is rounded
to the nearest, with ties rounded to even.
## Whats new
@ -96,25 +91,22 @@ The conversions supported cover the following cases.
[`Unwrapped`][u-1-8] wrappers:
* [`ZERO`][f-z-1-8], [`DELTA`][f-d-1-8]
* [`mul_acc`][f-ma-1-8]
* The [`ONE`][f-o-1-8] constant was added to all fixed-point numbers
that can represent the value 1.
* The following methods were added to all fixed-point numbers and to
the [`Fixed`][tf-1-8] trait:
* The [`ONE`][f-o-1-8] constant was added to all fixed-point numbers that can
represent the value 1.
* The following methods were added to all fixed-point numbers and to the
[`Fixed`][tf-1-8] trait:
* [`checked_mul_acc`][f-cma-1-8], [`saturating_mul_acc`][f-sma-1-8],
[`wrapping_mul_acc`][f-wma-1-8], [`unwrapped_mul_acc`][f-uma-1-8],
[`overflowing_mul_acc`][f-oma-1-8]
* [`saturating_div_euclid_int`][f-sdei-1-8],
[`saturating_rem_euclid_int`][f-srei-1-8]
* [`unwrapped_rem`][f-ur-1-8],
[`unwrapped_rem_euclid`][f-ure-1-8]
* [`unwrapped_rem`][f-ur-1-8], [`unwrapped_rem_euclid`][f-ure-1-8]
* [`unwrapped_rem_int`][f-uri-1-8]
* The following methods are now `const` functions:
* [`checked_rem`][f-cr-1-8]
* [`rem_euclid`][f-re-1-8], [`checked_rem_euclid`][f-cre-1-8]
* [`checked_div_int`][f-cdi-1-8],
[`wrapping_div_int`][f-wdi-1-8],
[`unwrapped_div_int`][f-udi-1-8],
[`overflowing_div_int`][f-odi-1-8]
* [`checked_div_int`][f-cdi-1-8], [`wrapping_div_int`][f-wdi-1-8],
[`unwrapped_div_int`][f-udi-1-8], [`overflowing_div_int`][f-odi-1-8]
* The following methods were added to all fixed-point numbers:
* [`const_not`][f-cn-1-8]
* [`const_bitand`][f-cba-1-8], [`const_bitor`][f-cbo-1-8],
@ -157,9 +149,9 @@ The conversions supported cover the following cases.
* [`next_power_of_two`][f-npot-1-7],
[`checked_next_power_of_two`][f-cnpot-1-7],
[`wrapping_next_power_of_two`][f-wnpot-1-7]
* The following constant and methods were added to all fixed-point
numbers, to the [`Fixed`][tf-1-7] trait, and to the
[`Wrapping`][w-1-7] and [`Unwrapped`][u-1-7] wrappers:
* The following constant and methods were added to all fixed-point numbers, to
the [`Fixed`][tf-1-7] trait, and to the [`Wrapping`][w-1-7] and
[`Unwrapped`][u-1-7] wrappers:
* [`IS_SIGNED`][f-is-1-7]
* [`from_be`][f-fb-1-7], [`from_le`][f-fl-1-7]
* [`to_be`][f-tb-1-7], [`to_le`][f-tl-1-7]
@ -172,36 +164,33 @@ The conversions supported cover the following cases.
[`from_ne_bytes`][f-fnb-1-7]
* [`to_be_bytes`][f-tbb-1-7], [`to_le_bytes`][f-tlb-1-7],
[`to_ne_bytes`][f-tnb-1-7]
* The following methods were added to all unsigned fixed-point
types, to the [`FixedUnsigned`][tfu-1-7] trait, and to the
[`Wrapping`][w-1-7] and [`Unwrapped`][u-1-7] wrappers for unsigned
numbers:
* The following methods were added to all unsigned fixed-point types, to the
[`FixedUnsigned`][tfu-1-7] trait, and to the [`Wrapping`][w-1-7] and
[`Unwrapped`][u-1-7] wrappers for unsigned numbers:
* [`significant_bits`][f-signi-1-7]
* [`highest_one`][f-ho-1-7]
* The [`signed_bits`][f-signe-1-7] method was added to all signed
fixed-point types, to the [`FixedSigned`][tfs-1-7] trait, and to
the [`Wrapping`][w-1-7] and [`Unwrapped`][u-1-7] wrappers for
signed numbers.
* The following constants, which are available in other programming
language standard libraries, were added to the [`consts`][c-1-7]
module and as associated constants to fixed-point types:
* The [`signed_bits`][f-signe-1-7] method was added to all signed fixed-point
types, to the [`FixedSigned`][tfs-1-7] trait, and to the [`Wrapping`][w-1-7]
and [`Unwrapped`][u-1-7] wrappers for signed numbers.
* The following constants, which are available in other programming language
standard libraries, were added to the [`consts`][c-1-7] module and as
associated constants to fixed-point types:
* [`SQRT_PI`][c-rp-1-7] (Go), [`FRAC_1_SQRT_PI`][c-1rp-1-7] (C++)
* [`SQRT_3`][c-r3-1-7] (C++), [`FRAC_1_SQRT_3`][c-1r3-1-7] (C++)
* [`SQRT_E`][c-re-1-7] (Go)
* [`SQRT_PHI`][c-rf-1-7] (Go)
* [`GAMMA`][c-g-1-7] (C++)
* [`CATALAN`][c-c-1-7] (Julia)
* [`Sum`] and [`Product`] are now supertraits of the
[`Fixed`][tf-1-7] trait.
* The [`F128Bits`][f128-1-7] type was added to support conversions
and comparisons between fixed-point numbers and *binary128*
floating-point numbers.
* [`Sum`] and [`Product`] are now supertraits of the [`Fixed`][tf-1-7] trait.
* The [`F128Bits`][f128-1-7] type was added to support conversions and
comparisons between fixed-point numbers and *binary128* floating-point
numbers.
* The features that previously required the [`az`][feat-dep-1-7] and
[`f16`][feat-dep-1-7] optional features are now always provided.
The [`az`][feat-dep-1-7] and [`f16`][feat-dep-1-7] optional
features are now deprecated and have no effect.
* For the experimental feature [`num-traits`][feat-exp-1-7], the
following traits were implemented for all fixed-point numbers:
[`f16`][feat-dep-1-7] optional features are now always provided. The
[`az`][feat-dep-1-7] and [`f16`][feat-dep-1-7] optional features are now
deprecated and have no effect.
* For the experimental feature [`num-traits`][feat-exp-1-7], the following
traits were implemented for all fixed-point numbers:
* [`OverflowingAdd`][nt-0-2-oa], [`OverflowingSub`][nt-0-2-os],
[`OverflowingMul`][nt-0-2-om]
@ -251,10 +240,9 @@ The conversions supported cover the following cases.
### Version 1.6.0 news (2021-02-05)
* The crate now requires rustc version 1.47.0 or later.
* The optional [*az* crate] dependency was updated to
[version 1.1][az-1-1].
* The [`unsigned_abs`][f-ua-1-6] method was added to all signed
fixed-point types and to the [`FixedSigned`][tfs-1-6] trait.
* The optional [*az* crate] dependency was updated to [version 1.1][az-1-1].
* The [`unsigned_abs`][f-ua-1-6] method was added to all signed fixed-point
types and to the [`FixedSigned`][tfs-1-6] trait.
* The following methods are now `const` functions:
* [`checked_neg`][f-cn-1-6], [`checked_add`][f-cad-1-6],
[`checked_sub`][f-cs-1-6], [`checked_mul_int`][f-cmi-1-6],
@ -302,12 +290,11 @@ assert_eq!(six_and_third.ceil().to_num::<i32>(), 7);
assert_eq!(six_and_third.ceil(), 7);
```
The type [`I20F12`] is a 32-bit fixed-point signed number with 20
integer bits and 12 fractional bits. It is an alias to
<code>[FixedI32][`FixedI32`]&lt;[U12][`U12`]&gt;</code>. The unsigned
counterpart would be [`U20F12`]. Aliases are provided for all
combinations of integer and fractional bits adding up to a total of
eight, 16, 32, 64 or 128 bits.
The type [`I20F12`] is a 32-bit fixed-point signed number with 20 integer bits
and 12 fractional bits. It is an alias to <code>[FixedI32]\<[U12]></code>. The
unsigned counterpart would be [`U20F12`]. Aliases are provided for all
combinations of integer and fractional bits adding up to a total of eight, 16,
32, 64 or 128 bits.
```rust
use fixed::types::{I4F4, I4F12};
@ -329,23 +316,22 @@ assert_eq!(ans2, I4F4::from_bits((3 << 4) + 6));
assert_eq!(ans2.to_string(), "3.4");
```
The second example shows some precision and conversion issues. The low
precision of `a` means that `a / 5` is 316 instead of 15, leading to
an inaccurate result `ans1` = 3 316 (~3.2). With a higher precision,
we get `wider_a / 5` equal to 8194096, leading to a more accurate
intermediate result `wider_ans` = 3 16354096. When we convert back to
four fractional bits, we get `ans2` = 3 616 (~3.4).
The second example shows some precision and conversion issues. The low precision
of `a` means that `a / 5` is 316 instead of 15, leading to an inaccurate
result `ans1` = 3 316 (~3.2). With a higher precision, we get `wider_a / 5`
equal to 8194096, leading to a more accurate intermediate result `wider_ans` =
3 16354096. When we convert back to four fractional bits, we get `ans2` = 3
616 (~3.4).
Note that we can convert from [`I4F4`] to [`I4F12`] using [`From`], as
the target type has the same number of integer bits and a larger
number of fractional bits. Converting from [`I4F12`] to [`I4F4`]
cannot use [`From`] as we have less fractional bits, so we use
[`from_num`] instead.
Note that we can convert from [`I4F4`] to [`I4F12`] using [`From`], as the
target type has the same number of integer bits and a larger number of
fractional bits. Converting from [`I4F12`] to [`I4F4`] cannot use [`From`] as we
have less fractional bits, so we use [`from_num`] instead.
## Writing fixed-point constants and values literally
The [*fixed-macro* crate] provides a convenient macro to write down
fixed-point constants literally in the code.
The [*fixed-macro* crate] provides a convenient macro to write down fixed-point
constants literally in the code.
```rust
use fixed::types::I16F16;
@ -358,8 +344,8 @@ assert_eq!(num2, 25.875);
## Using the *fixed* crate
The *fixed* crate is available on [crates.io][*fixed* crate]. To use
it in your crate, add it as a dependency inside [*Cargo.toml*]:
The *fixed* crate is available on [crates.io][*fixed* crate]. To use it in your
crate, add it as a dependency inside [*Cargo.toml*]:
```toml
[dependencies]
@ -372,21 +358,18 @@ The *fixed* crate requires rustc version 1.50.0 or later.
The *fixed* crate has these optional feature:
1. `serde`, disabled by default. This provides serialization support
for the fixed-point types. This feature requires the
[*serde* crate].
2. `std`, disabled by default. This is for features that are not
possible under `no_std`: currently the implementation of the
[`Error`] trait for [`ParseFixedError`].
3. `serde-str`, disabled by default. Fixed-point numbers are
serialized as strings showing the value when using human-readable
formats. This feature requires the `serde` and the `std` optional
features. **Warning:** numbers serialized when this feature is
enabled cannot be deserialized when this feature is disabled, and
vice versa.
1. `serde`, disabled by default. This provides serialization support for the
fixed-point types. This feature requires the [*serde* crate].
2. `std`, disabled by default. This is for features that are not possible under
`no_std`: currently the implementation of the [`Error`] trait for
[`ParseFixedError`].
3. `serde-str`, disabled by default. Fixed-point numbers are serialized as
strings showing the value when using human-readable formats. This feature
requires the `serde` and the `std` optional features. **Warning:** numbers
serialized when this feature is enabled cannot be deserialized when this
feature is disabled, and vice versa.
To enable features, you can add the dependency like this to
[*Cargo.toml*]:
To enable features, you can add the dependency like this to [*Cargo.toml*]:
```toml
[dependencies.fixed]
@ -396,33 +379,29 @@ features = ["serde"]
## Experimental optional features
It is not considered a breaking change if the following experimental
features are removed. The removal of experimental features would
however require a minor version bump. Similarly, on a minor version
bump, optional dependencies can be updated to an incompatible newer
version.
It is not considered a breaking change if the following experimental features
are removed. The removal of experimental features would however require a minor
version bump. Similarly, on a minor version bump, optional dependencies can be
updated to an incompatible newer version.
1. `num-traits`, disabled by default. This implements some traits
from the [*num-traits* crate]. (The plan is to promote this to an
optional feature once the [*num-traits* crate] reaches version
1.0.0.)
1. `num-traits`, disabled by default. This implements some traits from the
[*num-traits* crate]. (The plan is to promote this to an optional feature
once the [*num-traits* crate] reaches version 1.0.0.)
## Deprecated optional features
The following optional features are deprecated and may be removed in
the next major version of the crate.
The following optional features are deprecated and may be removed in the next
major version of the crate.
1. `az`, has no effect. Previously required to enable the cast traits
provided by the [*az* crate]. Now these cast traits are always
provided.
2. `f16`, has no effect. Previously required to provide conversion
to/from [`f16`] and [`bf16`]. Now these conversions are always
provided.
1. `az`, has no effect. Previously required to enable the cast traits provided
by the [*az* crate]. Now these cast traits are always provided.
2. `f16`, has no effect. Previously required to provide conversion to/from
[`f16`] and [`bf16`]. Now these conversions are always provided.
## License
This crate is free software: you can redistribute it and/or modify it
under the terms of either
This crate is free software: you can redistribute it and/or modify it under the
terms of either
* the [Apache License, Version 2.0][LICENSE-APACHE] or
* the [MIT License][LICENSE-MIT]
@ -431,10 +410,9 @@ at your option.
### Contribution
Unless you explicitly state otherwise, any contribution intentionally
submitted for inclusion in the work by you, as defined in the Apache
License, Version 2.0, shall be dual licensed as above, without any
additional terms or conditions.
Unless you explicitly state otherwise, any contribution intentionally submitted
for inclusion in the work by you, as defined in the Apache License, Version 2.0,
shall be dual licensed as above, without any additional terms or conditions.
[*Cargo.toml*]: https://doc.rust-lang.org/cargo/guide/dependencies.html
[*az* crate]: https://crates.io/crates/az
@ -452,6 +430,7 @@ additional terms or conditions.
[LICENSE-APACHE]: https://www.apache.org/licenses/LICENSE-2.0
[LICENSE-MIT]: https://opensource.org/licenses/MIT
[U0]: https://docs.rs/fixed/~1.7/fixed/types/extra/type.U0.html
[U12]: https://docs.rs/fixed/~1.7/fixed/types/extra/type.U12.html
[U32]: https://docs.rs/fixed/~1.7/fixed/types/extra/type.U32.html
[`Binary`]: https://doc.rust-lang.org/nightly/core/fmt/trait.Binary.html
[`Display`]: https://doc.rust-lang.org/nightly/core/fmt/trait.Display.html
@ -481,7 +460,6 @@ additional terms or conditions.
[`Octal`]: https://doc.rust-lang.org/nightly/core/fmt/trait.Octal.html
[`ParseFixedError`]: https://docs.rs/fixed/~1.7/fixed/struct.ParseFixedError.html
[`ToFixed`]: https://docs.rs/fixed/~1.7/fixed/traits/trait.ToFixed.html
[`U12`]: https://docs.rs/fixed/~1.7/fixed/types/extra/type.U12.html
[`U20F12`]: https://docs.rs/fixed/~1.7/fixed/types/type.U20F12.html
[`UpperHex`]: https://doc.rust-lang.org/nightly/core/fmt/trait.UpperHex.html
[`bf16`]: https://docs.rs/half/^1/half/struct.bf16.html

View File

@ -8,30 +8,27 @@ as-is, without any warranty. -->
Version 1.8.0 (unreleased)
==========================
* The following constants were added to all fixed-point numbers, to
the [`Fixed`][tf-1-8] trait, and to the [`Wrapping`][w-1-8] and
* The following constants were added to all fixed-point numbers, to the
[`Fixed`][tf-1-8] trait, and to the [`Wrapping`][w-1-8] and
[`Unwrapped`][u-1-8] wrappers:
* [`ZERO`][f-z-1-8], [`DELTA`][f-d-1-8]
* [`mul_acc`][f-ma-1-8]
* The [`ONE`][f-o-1-8] constant was added to all fixed-point numbers
that can represent the value 1.
* The following methods were added to all fixed-point numbers and to
the [`Fixed`][tf-1-8] trait:
* The [`ONE`][f-o-1-8] constant was added to all fixed-point numbers that can
represent the value 1.
* The following methods were added to all fixed-point numbers and to the
[`Fixed`][tf-1-8] trait:
* [`checked_mul_acc`][f-cma-1-8], [`saturating_mul_acc`][f-sma-1-8],
[`wrapping_mul_acc`][f-wma-1-8], [`unwrapped_mul_acc`][f-uma-1-8],
[`overflowing_mul_acc`][f-oma-1-8]
* [`saturating_div_euclid_int`][f-sdei-1-8],
[`saturating_rem_euclid_int`][f-srei-1-8]
* [`unwrapped_rem`][f-ur-1-8],
[`unwrapped_rem_euclid`][f-ure-1-8]
* [`unwrapped_rem`][f-ur-1-8], [`unwrapped_rem_euclid`][f-ure-1-8]
* [`unwrapped_rem_int`][f-uri-1-8]
* The following methods are now `const` functions:
* [`checked_rem`][f-cr-1-8]
* [`rem_euclid`][f-re-1-8], [`checked_rem_euclid`][f-cre-1-8]
* [`checked_div_int`][f-cdi-1-8],
[`wrapping_div_int`][f-wdi-1-8],
[`unwrapped_div_int`][f-udi-1-8],
[`overflowing_div_int`][f-odi-1-8]
* [`checked_div_int`][f-cdi-1-8], [`wrapping_div_int`][f-wdi-1-8],
[`unwrapped_div_int`][f-udi-1-8], [`overflowing_div_int`][f-odi-1-8]
* The following methods were added to all fixed-point numbers:
* [`const_not`][f-cn-1-8]
* [`const_bitand`][f-cba-1-8], [`const_bitor`][f-cbo-1-8],
@ -75,9 +72,9 @@ Version 1.7.0 (2021-03-25)
* [`next_power_of_two`][f-npot-1-7],
[`checked_next_power_of_two`][f-cnpot-1-7],
[`wrapping_next_power_of_two`][f-wnpot-1-7]
* The following constant and methods were added to all fixed-point
numbers, to the [`Fixed`][tf-1-7] trait, and to the
[`Wrapping`][w-1-7] and [`Unwrapped`][u-1-7] wrappers:
* The following constant and methods were added to all fixed-point numbers, to
the [`Fixed`][tf-1-7] trait, and to the [`Wrapping`][w-1-7] and
[`Unwrapped`][u-1-7] wrappers:
* [`IS_SIGNED`][f-is-1-7]
* [`from_be`][f-fb-1-7], [`from_le`][f-fl-1-7]
* [`to_be`][f-tb-1-7], [`to_le`][f-tl-1-7]
@ -90,36 +87,33 @@ Version 1.7.0 (2021-03-25)
[`from_ne_bytes`][f-fnb-1-7]
* [`to_be_bytes`][f-tbb-1-7], [`to_le_bytes`][f-tlb-1-7],
[`to_ne_bytes`][f-tnb-1-7]
* The following methods were added to all unsigned fixed-point
types, to the [`FixedUnsigned`][tfu-1-7] trait, and to the
[`Wrapping`][w-1-7] and [`Unwrapped`][u-1-7] wrappers for unsigned
numbers:
* The following methods were added to all unsigned fixed-point types, to the
[`FixedUnsigned`][tfu-1-7] trait, and to the [`Wrapping`][w-1-7] and
[`Unwrapped`][u-1-7] wrappers for unsigned numbers:
* [`significant_bits`][f-signi-1-7]
* [`highest_one`][f-ho-1-7]
* The [`signed_bits`][f-signe-1-7] method was added to all signed
fixed-point types, to the [`FixedSigned`][tfs-1-7] trait, and to
the [`Wrapping`][w-1-7] and [`Unwrapped`][u-1-7] wrappers for
signed numbers.
* The following constants, which are available in other programming
language standard libraries, were added to the [`consts`][c-1-7]
module and as associated constants to fixed-point types:
* The [`signed_bits`][f-signe-1-7] method was added to all signed fixed-point
types, to the [`FixedSigned`][tfs-1-7] trait, and to the [`Wrapping`][w-1-7]
and [`Unwrapped`][u-1-7] wrappers for signed numbers.
* The following constants, which are available in other programming language
standard libraries, were added to the [`consts`][c-1-7] module and as
associated constants to fixed-point types:
* [`SQRT_PI`][c-rp-1-7] (Go), [`FRAC_1_SQRT_PI`][c-1rp-1-7] (C++)
* [`SQRT_3`][c-r3-1-7] (C++), [`FRAC_1_SQRT_3`][c-1r3-1-7] (C++)
* [`SQRT_E`][c-re-1-7] (Go)
* [`SQRT_PHI`][c-rf-1-7] (Go)
* [`GAMMA`][c-g-1-7] (C++)
* [`CATALAN`][c-c-1-7] (Julia)
* [`Sum`] and [`Product`] are now supertraits of the
[`Fixed`][tf-1-7] trait.
* The [`F128Bits`][f128-1-7] type was added to support conversions
and comparisons between fixed-point numbers and *binary128*
floating-point numbers.
* [`Sum`] and [`Product`] are now supertraits of the [`Fixed`][tf-1-7] trait.
* The [`F128Bits`][f128-1-7] type was added to support conversions and
comparisons between fixed-point numbers and *binary128* floating-point
numbers.
* The features that previously required the [`az`][feat-dep-1-7] and
[`f16`][feat-dep-1-7] optional features are now always provided.
The [`az`][feat-dep-1-7] and [`f16`][feat-dep-1-7] optional
features are now deprecated and have no effect.
* For the experimental feature [`num-traits`][feat-exp-1-7], the
following traits were implemented for all fixed-point numbers:
[`f16`][feat-dep-1-7] optional features are now always provided. The
[`az`][feat-dep-1-7] and [`f16`][feat-dep-1-7] optional features are now
deprecated and have no effect.
* For the experimental feature [`num-traits`][feat-exp-1-7], the following
traits were implemented for all fixed-point numbers:
* [`OverflowingAdd`][nt-0-2-oa], [`OverflowingSub`][nt-0-2-os],
[`OverflowingMul`][nt-0-2-om]
@ -168,10 +162,9 @@ Version 1.6.0 (2021-02-05)
==========================
* The crate now requires rustc version 1.47.0 or later.
* The optional [*az* crate] dependency was updated to
[version 1.1][az-1-1].
* The [`unsigned_abs`][f-ua-1-6] method was added to all signed
fixed-point types and to the [`FixedSigned`][tfs-1-6] trait.
* The optional [*az* crate] dependency was updated to [version 1.1][az-1-1].
* The [`unsigned_abs`][f-ua-1-6] method was added to all signed fixed-point
types and to the [`FixedSigned`][tfs-1-6] trait.
* The following methods are now `const` functions:
* [`checked_neg`][f-cn-1-6], [`checked_add`][f-cad-1-6],
[`checked_sub`][f-cs-1-6], [`checked_mul_int`][f-cmi-1-6],
@ -200,21 +193,19 @@ Version 1.6.0 (2021-02-05)
Version 1.5.0 (2020-11-05)
==========================
* The [`wide_mul`][f-wm-1-5] method was added to all fixed-point
numbers up to 64 bits wide ([issue 25]).
* Unwrapped methods for arithmetic together with the
[`Unwrapped`][unw-1-5] wrapper were added. Unwrapped methods panic
on overflow, even when debug assertions are disabled, similar to
how wrapping methods will wrap around even when debug assertions
are enabled. (This was previously an experimental feature
`unwrapped`.)
* The [`serde-str`][feat-1-5] feature was added. (This was
previously an experimental feature.)
* For the experimental feature [`num-traits`][feat-exp-1-5], some
missing supertraits were added to
[`FixedOptionalFeatures`][tfof-1-5].
* Bug fix: multiplication of [`FixedI128`][i128-1-5] was panicking
when multiplying some large negative numbers ([issue 26]).
* The [`wide_mul`][f-wm-1-5] method was added to all fixed-point numbers up to
64 bits wide ([issue 25]).
* Unwrapped methods for arithmetic together with the [`Unwrapped`][unw-1-5]
wrapper were added. Unwrapped methods panic on overflow, even when debug
assertions are disabled, similar to how wrapping methods will wrap around
even when debug assertions are enabled. (This was previously an experimental
feature `unwrapped`.)
* The [`serde-str`][feat-1-5] feature was added. (This was previously an
experimental feature.)
* For the experimental feature [`num-traits`][feat-exp-1-5], some missing
supertraits were added to [`FixedOptionalFeatures`][tfof-1-5].
* Bug fix: multiplication of [`FixedI128`][i128-1-5] was panicking when
multiplying some large negative numbers ([issue 26]).
[i128-1-5]: https://docs.rs/fixed/~1.5/fixed/struct.FixedI128.html
[f-wm-1-5]: https://docs.rs/fixed/~1.5/fixed/struct.FixedI32.html#method.wide_mul
@ -231,21 +222,19 @@ Version 1.4.0 (2020-10-22)
* The following methods were added to all fixed-point types, to the
[`Fixed`][tf-1-4] trait, and to the [`Wrapping`][w-1-4] wrapper:
* [`recip`][f-rec-1-4], [`checked_recip`][f-crec-1-4],
[`saturating_recip`][f-srec-1-4],
[`wrapping_recip`][f-wrec-1-4],
[`saturating_recip`][f-srec-1-4], [`wrapping_recip`][f-wrec-1-4],
[`overflowing_recip`][f-orec-1-4]
* For the experimental feature [`num-traits`][feat-exp-1-4], the
following traits were implemented where applicable ([issue 23]):
* For the experimental feature [`num-traits`][feat-exp-1-4], the following
traits were implemented where applicable ([issue 23]):
* [`Num`][nt-0-2-num]
* [`Signed`][nt-0-2-signed], [`Unsigned`][nt-0-2-unsigned]
* [`Inv`][nt-0-2-inv]
* [`MulAdd`][nt-0-2-ma], [`MulAddAssign`][nt-0-2-maa]
* [`FloatConst`][nt-0-2-fc]
* [`ToPrimitive`][nt-0-2-tp], [`FromPrimitive`][nt-0-2-fp]
* For the experimental feature [`serde-str`][feat-exp-1-4],
serialization in human-readable formats was made more convenient
to write manually ([issue 24]). This makes it incompatible with
the version in 1.3.0.
* For the experimental feature [`serde-str`][feat-exp-1-4], serialization in
human-readable formats was made more convenient to write manually ([issue
24]). This makes it incompatible with the version in 1.3.0.
[f-crec-1-4]: https://docs.rs/fixed/~1.4/fixed/struct.FixedI32.html#method.checked_recip
[f-orec-1-4]: https://docs.rs/fixed/~1.4/fixed/struct.FixedI32.html#method.overflowing_recip
@ -270,21 +259,18 @@ Version 1.4.0 (2020-10-22)
Version 1.3.0 (2020-10-15)
==========================
* The [`MulAssign`] implementation on fixed-point numbers now
accepts an rhs fixed-point number with a different number of
fractional bits from `self`.
* The [`MulAssign`] implementation on fixed-point numbers now accepts an rhs
fixed-point number with a different number of fractional bits from `self`.
* The following methods were added to all fixed-point types, to the
[`Fixed`][tf-1-3] trait, and to the [`Wrapping`][w-1-3] wrapper:
* [`mul_add`][f-ma-1-3], [`checked_mul_add`][f-cma-1-3],
[`saturating_mul_add`][f-sma-1-3],
[`wrapping_mul_add`][f-wma-1-3],
[`saturating_mul_add`][f-sma-1-3], [`wrapping_mul_add`][f-wma-1-3],
[`overflowing_mul_add`][f-oma-1-3]
* The new experimental feature [`unwrapped`][feat-exp-1-3] was
added, providing arithmetic methods that panic on overflow even
when debug assertions are disabled.
* The new experimental feature [`serde-str`][feat-exp-1-3] was
added, which makes serialization use the numbers value in
human-readable formats.
* The new experimental feature [`unwrapped`][feat-exp-1-3] was added,
providing arithmetic methods that panic on overflow even when debug
assertions are disabled.
* The new experimental feature [`serde-str`][feat-exp-1-3] was added, which
makes serialization use the numbers value in human-readable formats.
[f-cma-1-3]: https://docs.rs/fixed/~1.3/fixed/struct.FixedI32.html#method.checked_mul_add
[f-ma-1-3]: https://docs.rs/fixed/~1.3/fixed/struct.FixedI32.html#method.mul_add
@ -298,9 +284,8 @@ Version 1.3.0 (2020-10-15)
Version 1.2.0 (2020-09-02)
==========================
* The [`const_fixed_from_int!`][cffi-1-2] macro was added to make it
easy to define constant fixed-point numbers using integer
expressions ([issue 20]).
* The [`const_fixed_from_int!`][cffi-1-2] macro was added to make it easy to
define constant fixed-point numbers using integer expressions ([issue 20]).
[cffi-1-2]: https://docs.rs/fixed/~1.2/fixed/macro.const_fixed_from_int.html
[issue 20]: https://gitlab.com/tspiteri/fixed/-/issues/20
@ -308,10 +293,9 @@ Version 1.2.0 (2020-09-02)
Version 1.1.0 (2020-07-21)
==========================
* The new experimental feature [`num-traits`][feat-nt-1-1] was added
to implement some traits, and to also add the relevant traits as
supertraits to the [`FixedOptionalFeatures`][fof-1-1] trait
([issue 18]).
* The new experimental feature [`num-traits`][feat-nt-1-1] was added to
implement some traits, and to also add the relevant traits as supertraits to
the [`FixedOptionalFeatures`][fof-1-1] trait ([issue 18]).
[issue 18]: https://gitlab.com/tspiteri/fixed/-/issues/18
[feat-nt-1-1]: https://docs.rs/fixed/~1.1/fixed/#experimental-optional-features
@ -339,20 +323,20 @@ Version 0.5.7 (2020-05-11)
==========================
* The `LosslessTryFrom` and `LosslessTryInto` traits were added.
* The following methods were added to all fixed-point types, to the
`Fixed` trait, and to the `Wrapping` wrapper:
* The following methods were added to all fixed-point types, to the `Fixed`
trait, and to the `Wrapping` wrapper:
* `leading_ones`, `trailing_ones`
* The following method was added to unsigned fixed-point types and
to the `FixedUnsigned` trait:
* The following method was added to unsigned fixed-point types and to the
`FixedUnsigned` trait:
* `wrapping_next_power_of_two`
* The `PHI` and `FRAC_1_PHI` constants were added to the `consts`
module and as associated constants to fixed-point types.
* The `PHI` and `FRAC_1_PHI` constants were added to the `consts` module and
as associated constants to fixed-point types.
Version 0.5.6 (2020-05-01)
==========================
* The following methods were added to signed fixed-point types and to
the `FixedSigned` trait:
* The following methods were added to signed fixed-point types and to the
`FixedSigned` trait:
* `checked_signum`, `saturating_signum`, `wrapping_signum`,
`overflowing_signum`
* The `LossyFrom` and `LossyInto` traits were added to the prelude.
@ -362,23 +346,23 @@ Version 0.5.6 (2020-05-01)
Version 0.5.5 (2020-04-16)
==========================
* Bug fix: an incorrect result could be given when comparing a
signed fixed-point number of type `FixedI` to a number that would
overflow by exactly one bit when converting to `FixedI`.
* The following associated constants were added to all fixed-point
types, to the `Fixed` trait, and to the `Wrapping` wrapper:
* Bug fix: an incorrect result could be given when comparing a signed
fixed-point number of type `FixedI` to a number that would overflow by
exactly one bit when converting to `FixedI`.
* The following associated constants were added to all fixed-point types, to
the `Fixed` trait, and to the `Wrapping` wrapper:
* `MIN`, `MAX`
* The following associated constants were added to the `Fixed` trait
and to the `Wrapping` wrapper:
* The following associated constants were added to the `Fixed` trait and to
the `Wrapping` wrapper:
* `INT_NBITS`, `FRAC_NBITS`
* The following methods were added to all fixed-point types and to
the `Fixed` trait:
* The following methods were added to all fixed-point types and to the `Fixed`
trait:
* `int_log2`, `int_log10`
* `checked_int_log2`, `checked_int_log10`
* The following methods were added to the `Wrapping` wrapper:
* `int_log2`, `int_log10`
* The constants in the `consts` module were also added as associated
constants to fixed-point types that can represent them.
* The constants in the `consts` module were also added as associated constants
to fixed-point types that can represent them.
* The following methods were deprecated:
* `min_value`, `max_value`
* `int_nbits`, `frac_nbits`
@ -386,22 +370,22 @@ Version 0.5.5 (2020-04-16)
Version 0.5.4 (2020-02-21)
==========================
* Bug fix: `rem_euclid_int` and its checked versions were handling
overflow incorrectly.
* Bug fix: `rem_euclid_int` and its checked versions were handling overflow
incorrectly.
Version 0.5.3 (2020-02-13)
==========================
* Bug fix: `round_to_zero` was returning incorrect results for
negative whole number operands.
* Bug fix: all remainder operations with a fixed-point LHS and an
integer RHS were giving an incorrect answer
* Bug fix: `round_to_zero` was returning incorrect results for negative whole
number operands.
* Bug fix: all remainder operations with a fixed-point LHS and an integer RHS
were giving an incorrect answer
(https://gitlab.com/tspiteri/fixed/issues/13).
* Bug fix: Euclidean division operations by integers were giving an
incorrect answer.
* Bug fix: Euclidean division operations by integers were giving an incorrect
answer.
* `Rem` and `RemAssign` were implemented for fixed-point numbers.
* The following methods were added to all fixed-point types and to
the `Fixed` trait:
* The following methods were added to all fixed-point types and to the `Fixed`
trait:
* `checked_rem`
* `div_euclid`, `rem_euclid`
* `checked_div_euclid`, `checked_rem_euclid`
@ -422,15 +406,14 @@ Version 0.5.2 (2020-02-02)
Version 0.5.1 (2019-12-22)
==========================
* `ParseFixedError` implements `Error` when the new `std` feature is
enabled.
* `ParseFixedError` implements `Error` when the new `std` feature is enabled.
Version 0.5.0 (2019-12-06)
==========================
* The crate now requires rustc version 1.39.0 or later.
* The following methods were added to all fixed-point types and to
the `Fixed` trait:
* The following methods were added to all fixed-point types and to the `Fixed`
trait:
* `from_be_bytes`, `from_le_bytes`, `from_ne_bytes`
* `to_be_bytes`, `to_le_bytes`, `to_ne_bytes`
* `div_euclid_int`, `rem_euclid_int`
@ -448,99 +431,92 @@ Incompatible changes
Version 0.4.6 (2019-10-16)
==========================
* Conversions to/from `bf16` are now provided when the `f16` option
is enabled.
* Conversions to/from `bf16` are now provided when the `f16` option is
enabled.
* The following methods are now `const` functions: `saturating_neg`,
`saturating_add`, `saturating_sub`, `saturating_mul_int`,
`saturating_abs`
`saturating_add`, `saturating_sub`, `saturating_mul_int`, `saturating_abs`
* Support for casts using the *az* crate was added.
Version 0.4.5 (2019-08-30)
==========================
* Bug fix: display of many decimal numbers was panicking in debug
mode or including a leading zero in release mode.
* Many methods were added to `Wrapping` for convenience, even if
they do not involve wrapping.
* Bug fix: display of many decimal numbers was panicking in debug mode or
including a leading zero in release mode.
* Many methods were added to `Wrapping` for convenience, even if they do not
involve wrapping.
Version 0.4.4 (2019-08-24)
==========================
* Bug fix: rounding could produce bad output for `Binary`, `Octal`,
`LowerHex` and `UpperHex`.
* The following methods are now `const` functions:
`is_power_of_two`, `abs`, `wrapping_abs`, `overflowing_abs`
* Bug fix: rounding could produce bad output for `Binary`, `Octal`, `LowerHex`
and `UpperHex`.
* The following methods are now `const` functions: `is_power_of_two`, `abs`,
`wrapping_abs`, `overflowing_abs`
* The method `round_to_zero` was added.
* The method `round_ties_to_even` and its checked versions were
added.
* The method `round_ties_to_even` and its checked versions were added.
Version 0.4.3 (2019-08-20)
==========================
* The crate now requires rustc version 1.34.0 or later.
* The precision argument is no longer ignored when formatting
fixed-point numbers; precision should now be handled the same as
for primitive floating-point numbers in the standard library.
* Parsing strings now rounds to the nearest with ties rounded to
even.
* Checked versions of string parsing methods are now available as
inherent methods to all fixed-point numbers, and as methods in the
`Fixed` trait.
* `Wrapping` now has methods for parsing with wrapping, including
an implementation of `FromStr`.
* The precision argument is no longer ignored when formatting fixed-point
numbers; precision should now be handled the same as for primitive
floating-point numbers in the standard library.
* Parsing strings now rounds to the nearest with ties rounded to even.
* Checked versions of string parsing methods are now available as inherent
methods to all fixed-point numbers, and as methods in the `Fixed` trait.
* `Wrapping` now has methods for parsing with wrapping, including an
implementation of `FromStr`.
* The following methods are now `const` functions:
* `min_value`, `max_value`, `from_bits`, `to_bits`
* `count_ones`, `count_zeros`, `leading_zeros`, `trailing_zeros`
`rotate_left`, `rotate_right`
* `wrapping_neg`, `wrapping_add`, `wrapping_sub`,
`wrapping_mul_int`, `wrapping_shl`, `wrapping_shr`
* `wrapping_neg`, `wrapping_add`, `wrapping_sub`, `wrapping_mul_int`,
`wrapping_shl`, `wrapping_shr`
* `overflowing_neg`, `overflowing_add`, `overflowing_sub`,
`overflowing_mul_int`, `overflowing_shl`, `overflowing_shr`
* `is_positive`, `is_negative`
* The associated constants `INT_NBITS` and `FRAC_NBITS` were added.
* The reexports in the `frac` module and the `LeEqU*` traits were
moved into the new `types::extra` module.
* The reexports in the `frac` module and the `LeEqU*` traits were moved into
the new `types::extra` module.
Version 0.4.2 (2019-08-16)
==========================
* The new methods `from_num` and `to_num` together with their
checked versions were added to all fixed-point numbers.
* The methods `from_fixed`, `to_fixed`, `from_int`, `to_int`,
`from_float`, and `to_float`, and their checked versions, were
deprecated.
* The new methods `from_num` and `to_num` together with their checked versions
were added to all fixed-point numbers.
* The methods `from_fixed`, `to_fixed`, `from_int`, `to_int`, `from_float`,
and `to_float`, and their checked versions, were deprecated.
* The new method `from_num` was added to the `Wrapping` wrapper.
* Bug fix: parsing of decimal fractions was fixed to give correctly
rounded results for long decimal fraction strings, for example
with four fractional bits, 0.96874999… (just below 3132) and
0.96875 (3132) are now parsed correctly as 0.9375 (1516) and 1.0.
* Bug fix: parsing of decimal fractions was fixed to give correctly rounded
results for long decimal fraction strings, for example with four fractional
bits, 0.96874999… (just below 3132) and 0.96875 (3132) are now parsed
correctly as 0.9375 (1516) and 1.0.
Version 0.4.1 (2019-08-12)
==========================
* All fixed-point types now implement `FromStr`.
* The methods `from_str_binary`, `from_str_octal` and `from_str_hex`
were added.
* The methods `from_str_binary`, `from_str_octal` and `from_str_hex` were
added.
Version 0.4.0 (2019-08-08)
==========================
* The crate now requires rustc version 1.31.0 or later.
* The `traits` module was added, with its traits `Fixed`,
`FixedSigned`, `FixedUnsigned`, `FromFixed`, `ToFixed`,
`LossyFrom` and `LossyInto`.
* The `saturating_neg` method was added to all fixed-point numbers,
and the `saturating_abs` method was added to signed fixed-point
numbers.
* The `traits` module was added, with its traits `Fixed`, `FixedSigned`,
`FixedUnsigned`, `FromFixed`, `ToFixed`, `LossyFrom` and `LossyInto`.
* The `saturating_neg` method was added to all fixed-point numbers, and the
`saturating_abs` method was added to signed fixed-point numbers.
* The `consts` module was added.
* The `signum` method now wraps instead of panics in release mode.
Incompatible changes
--------------------
* The sealed traits `Int` and `Float` now have no provided methods;
the methods in the old implementation are now provided by
`FromFixed` and `ToFixed`.
* The sealed traits `Int` and `Float` now have no provided methods; the
methods in the old implementation are now provided by `FromFixed` and
`ToFixed`.
* Deprecated methods were removed.
Contributors
@ -562,65 +538,57 @@ Version 0.3.2 (2019-02-27)
Version 0.3.1 (2019-02-07)
==========================
* Reimplement `From<bool>` for all fixed-point types which can
represent the integer 1. This was inadvertently removed in 0.3.0.
* Reimplement `From<bool>` for all fixed-point types which can represent the
integer 1. This was inadvertently removed in 0.3.0.
Version 0.3.0 (2019-02-03)
==========================
* Incompatible change: the return type of `to_int` is now generic.
* Incompatible change: the `Int` trait implementation for `bool` was
removed.
* Incompatible change: the `Int` trait implementation for `bool` was removed.
* The new method `to_fixed` was added.
* The new methods `checked_to_fixed`, `checked_to_int`,
`saturating_to_fixed`, `saturating_to_int`, `wrapping_to_fixed`,
`wrapping_to_int`, `overflowing_to_fixed` and `overflowing_to_int`
were added.
* The new methods `checked_to_fixed`, `checked_to_int`, `saturating_to_fixed`,
`saturating_to_int`, `wrapping_to_fixed`, `wrapping_to_int`,
`overflowing_to_fixed` and `overflowing_to_int` were added.
* The methods `from_fixed`, `to_fixed`, `checked_from_fixed`,
`checked_to_fixed`, `saturating_from_fixed`,
`saturating_to_fixed`, `wrapping_from_fixed`, `wrapping_to_fixed`,
`overflowing_from_fixed` and `overflowing_to_fixed` were added to
the `Int` trait.
`checked_to_fixed`, `saturating_from_fixed`, `saturating_to_fixed`,
`wrapping_from_fixed`, `wrapping_to_fixed`, `overflowing_from_fixed` and
`overflowing_to_fixed` were added to the `Int` trait.
* The methods `from_fixed`, `to_fixed`, `checked_to_fixed`,
`saturating_to_fixed`, `wrapping_to_fixed` and
`overflowing_to_fixed` were added to the `Float` trait.
* `PartialEq` and `PartialCmp` are now implemented for all
combinations of fixed-point numbers and primitive integers.
* The methods `int_bits` and `frac_bits` were deprecated and
replaced by the methods `int_nbits` and `frac_nbits`.
`saturating_to_fixed`, `wrapping_to_fixed` and `overflowing_to_fixed` were
added to the `Float` trait.
* `PartialEq` and `PartialCmp` are now implemented for all combinations of
fixed-point numbers and primitive integers.
* The methods `int_bits` and `frac_bits` were deprecated and replaced by the
methods `int_nbits` and `frac_nbits`.
Version 0.2.1 (2019-01-29)
==========================
* Bug fix: the `from_fixed` and `from_int` methods (and their
checked counterparts) could return wrong values for negative
values.
* Bug fix: display was using one fractional digit less than
required, thus yielding the same output for diffent fixed-point
numbers.
* Bug fix: the `from_fixed` and `from_int` methods (and their checked
counterparts) could return wrong values for negative values.
* Bug fix: display was using one fractional digit less than required, thus
yielding the same output for diffent fixed-point numbers.
Version 0.2.0 (2019-01-29)
==========================
* Incompatible change: The method `from_int` was change to accept a
generic parameter.
* The new methods `from_fixed`, `checked_from_fixed`,
`saturating_from_fixed`, `wrapping_from_fixed` and
`overflowing_from_fixed` were added.
* Incompatible change: The method `from_int` was change to accept a generic
parameter.
* The new methods `from_fixed`, `checked_from_fixed`, `saturating_from_fixed`,
`wrapping_from_fixed` and `overflowing_from_fixed` were added.
* The new methods `checked_from_int`, `saturating_from_int`,
`wrapping_from_int` and `overflowing_from_int` were added.
* The new methods `from_float`, `checked_from_float`,
`saturating_from_float`, `wrapping_from_float` and
`overflowing_from_float` were added.
* The new methods `from_float`, `checked_from_float`, `saturating_from_float`,
`wrapping_from_float` and `overflowing_from_float` were added.
* The new method `to_float` was added.
* The methods `from_f16`, `from_f32`, `from_f64`, `to_f16`, `to_f32`
and `to_f64` were deprecated.
* The `to_int` method was fixed to truncate fractional bits as
documented for negative values.
* The new methods `ceil`, `floor`, `round`, `checked_ceil`,
`checked_floor`, `checked_round`, `saturating_ceil`,
`saturating_floor`, `saturating_round`, `wrapping_ceil`,
`wrapping_floor`, `wrapping_round`, `overflowing_ceil`,
* The methods `from_f16`, `from_f32`, `from_f64`, `to_f16`, `to_f32` and
`to_f64` were deprecated.
* The `to_int` method was fixed to truncate fractional bits as documented for
negative values.
* The new methods `ceil`, `floor`, `round`, `checked_ceil`, `checked_floor`,
`checked_round`, `saturating_ceil`, `saturating_floor`, `saturating_round`,
`wrapping_ceil`, `wrapping_floor`, `wrapping_round`, `overflowing_ceil`,
`overflowing_floor` and `overflowing_round` were added.
* The methods `to_int_ceil`, `to_int_floor` and `to_int_round` were
deprecated.
@ -633,10 +601,10 @@ Version 0.1.6 (2019-01-27)
Version 0.1.5 (2019-01-26)
==========================
* Lossless infallible conversions between fixed-point numbers and
numeric primitives are now supported using `From` and `Into`.
* A new module `types` is available with aliases for all supported
fixed-point numbers.
* Lossless infallible conversions between fixed-point numbers and numeric
primitives are now supported using `From` and `Into`.
* A new module `types` is available with aliases for all supported fixed-point
numbers.
Version 0.1.4 (2018-11-29)
==========================
@ -646,35 +614,35 @@ Version 0.1.4 (2018-11-29)
Version 0.1.3 (2018-08-23)
==========================
* The `f16` feature was added, and new methods `from_f16` and
`to_f16` were added.
* The `f16` feature was added, and new methods `from_f16` and `to_f16` were
added.
Version 0.1.2 (2018-08-15)
==========================
* The crate can now be used without the standard library `std`.
* New methods `from_f32` and `from_f64` were added.
* New methods `is_positive` and `is_negative` were added to signed
fixed-point numbers.
* New methods `is_positive` and `is_negative` were added to signed fixed-point
numbers.
Version 0.1.1 (2018-08-11)
==========================
* Comparisons are now supported between all fixed-point numbers with
the same underlying integer type.
* Comparisons are now supported between all fixed-point numbers with the same
underlying integer type.
* New static methods `int_bits` and `frac_bits` were added.
* New methods `from_int`, `to_int`, `to_int_ceil`, `to_int_floor`
and `to_int_round` were added.
* New methods `from_int`, `to_int`, `to_int_ceil`, `to_int_floor` and
`to_int_round` were added.
* New methods `int` and `frac` were added.
* Support for multiplication and division by integers was added.
Version 0.1.0 (2018-08-10)
==========================
* `Unsigned` constants provided by the *typenum* crate are now used
for the number of fractional bits.
* Many methods and trait implementations available for primitive
integers are now also supported by the fixed-point numbers.
* `Unsigned` constants provided by the *typenum* crate are now used for the
number of fractional bits.
* Many methods and trait implementations available for primitive integers are
now also supported by the fixed-point numbers.
[*az* crate]: https://crates.io/crates/az
[`MulAssign`]: https://doc.rust-lang.org/nightly/core/ops/trait.MulAssign.html

View File

@ -39,10 +39,9 @@ range 0.5 ≤ <i>x</i> < 0.5 for signed numbers like
<code>[FixedI32]\<[U32]></code>, and in the range 0  <i>x</i> < 1 for unsigned
numbers like <code>[FixedU32]\<[U32]></code>.
In version 1 the [*typenum* crate] is used for the fractional bit
count `Frac`; the plan is to to have a major version 2 with [const
generics] instead when the Rust compiler support for them is powerful
enough.
In version 1 the [*typenum* crate] is used for the fractional bit count `Frac`;
the plan is to to have a major version 2 with [const generics] instead when the
Rust compiler support for them is powerful enough.
The main features are
@ -55,46 +54,42 @@ The main features are
This crate does *not* provide general analytic functions.
* No algebraic functions are provided, for example no `sqrt` or
`pow`.
* No trigonometric functions are provided, for example no `sin` or
`cos`.
* No other transcendental functions are provided, for example no
`log` or `exp`.
* No algebraic functions are provided, for example no `sqrt` or `pow`.
* No trigonometric functions are provided, for example no `sin` or `cos`.
* No other transcendental functions are provided, for example no `log` or
`exp`.
These functions are not provided because different implementations can
have different trade-offs, for example trading some correctness for
speed. Implementations can be provided in other crates.
These functions are not provided because different implementations can have
different trade-offs, for example trading some correctness for speed.
Implementations can be provided in other crates.
* The [*fixed-sqrt* crate] provides the square root operation.
* The [*cordic* crate] provides various functions implemented using
the [CORDIC] algorithm.
* The [*cordic* crate] provides various functions implemented using the
[CORDIC] algorithm.
The conversions supported cover the following cases.
* Infallible lossless conversions between fixed-point numbers and
numeric primitives are provided using [`From`] and [`Into`]. These
never fail (infallible) and do not lose any bits (lossless).
* Infallible lossy conversions between fixed-point numbers and
numeric primitives are provided using the [`LossyFrom`] and
[`LossyInto`] traits. The source can have more fractional bits
than the destination.
* Checked lossless conversions between fixed-point numbers and
numeric primitives are provided using the [`LosslessTryFrom`] and
[`LosslessTryInto`] traits. The source cannot have more fractional
bits than the destination.
* Checked conversions between fixed-point numbers and numeric
primitives are provided using the [`FromFixed`] and [`ToFixed`]
traits, or using the [`from_num`] and [`to_num`] methods and
[their checked versions][`checked_from_num`].
* Fixed-point numbers can be parsed from decimal strings using
[`FromStr`], and from binary, octal and hexadecimal strings using
the [`from_str_binary`], [`from_str_octal`] and [`from_str_hex`]
methods. The result is rounded to the nearest, with ties rounded
to even.
* Infallible lossless conversions between fixed-point numbers and numeric
primitives are provided using [`From`] and [`Into`]. These never fail
(infallible) and do not lose any bits (lossless).
* Infallible lossy conversions between fixed-point numbers and numeric
primitives are provided using the [`LossyFrom`] and [`LossyInto`] traits.
The source can have more fractional bits than the destination.
* Checked lossless conversions between fixed-point numbers and numeric
primitives are provided using the [`LosslessTryFrom`] and
[`LosslessTryInto`] traits. The source cannot have more fractional bits than
the destination.
* Checked conversions between fixed-point numbers and numeric primitives are
provided using the [`FromFixed`] and [`ToFixed`] traits, or using the
[`from_num`] and [`to_num`] methods and [their checked
versions][`checked_from_num`].
* Fixed-point numbers can be parsed from decimal strings using [`FromStr`],
and from binary, octal and hexadecimal strings using the
[`from_str_binary`], [`from_str_octal`] and [`from_str_hex`] methods. The
result is rounded to the nearest, with ties rounded to even.
* Fixed-point numbers can be converted to strings using [`Display`],
[`Binary`], [`Octal`], [`LowerHex`] and [`UpperHex`]. The output
is rounded to the nearest, with ties rounded to even.
[`Binary`], [`Octal`], [`LowerHex`] and [`UpperHex`]. The output is rounded
to the nearest, with ties rounded to even.
## Quick examples
@ -111,12 +106,11 @@ assert_eq!(six_and_third.ceil().to_num::<i32>(), 7);
assert_eq!(six_and_third.ceil(), 7);
```
The type [`I20F12`] is a 32-bit fixed-point signed number with 20
integer bits and 12 fractional bits. It is an alias to
<code>[FixedI32][`FixedI32`]&lt;[U12][`U12`]&gt;</code>. The unsigned
counterpart would be [`U20F12`]. Aliases are provided for all
combinations of integer and fractional bits adding up to a total of
eight, 16, 32, 64 or 128 bits.
The type [`I20F12`] is a 32-bit fixed-point signed number with 20 integer bits
and 12 fractional bits. It is an alias to <code>[FixedI32]\<[U12]></code>. The
unsigned counterpart would be [`U20F12`]. Aliases are provided for all
combinations of integer and fractional bits adding up to a total of eight, 16,
32, 64 or 128 bits.
```rust
use fixed::types::{I4F4, I4F12};
@ -138,23 +132,22 @@ assert_eq!(ans2, I4F4::from_bits((3 << 4) + 6));
assert_eq!(ans2.to_string(), "3.4");
```
The second example shows some precision and conversion issues. The low
precision of `a` means that `a / 5` is 316 instead of 15, leading to
an inaccurate result `ans1` = 3 316 (~3.2). With a higher precision,
we get `wider_a / 5` equal to 8194096, leading to a more accurate
intermediate result `wider_ans` = 3 16354096. When we convert back to
four fractional bits, we get `ans2` = 3 616 (~3.4).
The second example shows some precision and conversion issues. The low precision
of `a` means that `a / 5` is 316 instead of 15, leading to an inaccurate
result `ans1` = 3 316 (~3.2). With a higher precision, we get `wider_a / 5`
equal to 8194096, leading to a more accurate intermediate result `wider_ans` =
3 16354096. When we convert back to four fractional bits, we get `ans2` = 3
616 (~3.4).
Note that we can convert from [`I4F4`] to [`I4F12`] using [`From`], as
the target type has the same number of integer bits and a larger
number of fractional bits. Converting from [`I4F12`] to [`I4F4`]
cannot use [`From`] as we have less fractional bits, so we use
[`from_num`] instead.
Note that we can convert from [`I4F4`] to [`I4F12`] using [`From`], as the
target type has the same number of integer bits and a larger number of
fractional bits. Converting from [`I4F12`] to [`I4F4`] cannot use [`From`] as we
have less fractional bits, so we use [`from_num`] instead.
## Writing fixed-point constants and values literally
The [*fixed-macro* crate] provides a convenient macro to write down
fixed-point constants literally in the code.
The [*fixed-macro* crate] provides a convenient macro to write down fixed-point
constants literally in the code.
```rust
# #[cfg(feature = "skip-this-test")] {
@ -169,8 +162,8 @@ assert_eq!(num2, 25.875);
## Using the *fixed* crate
The *fixed* crate is available on [crates.io][*fixed* crate]. To use
it in your crate, add it as a dependency inside [*Cargo.toml*]:
The *fixed* crate is available on [crates.io][*fixed* crate]. To use it in your
crate, add it as a dependency inside [*Cargo.toml*]:
```toml
[dependencies]
@ -183,21 +176,18 @@ The *fixed* crate requires rustc version 1.50.0 or later.
The *fixed* crate has these optional feature:
1. `serde`, disabled by default. This provides serialization support
for the fixed-point types. This feature requires the
[*serde* crate].
2. `std`, disabled by default. This is for features that are not
possible under `no_std`: currently the implementation of the
[`Error`] trait for [`ParseFixedError`].
3. `serde-str`, disabled by default. Fixed-point numbers are
serialized as strings showing the value when using human-readable
formats. This feature requires the `serde` and the `std` optional
features. **Warning:** numbers serialized when this feature is
enabled cannot be deserialized when this feature is disabled, and
vice versa.
1. `serde`, disabled by default. This provides serialization support for the
fixed-point types. This feature requires the [*serde* crate].
2. `std`, disabled by default. This is for features that are not possible under
`no_std`: currently the implementation of the [`Error`] trait for
[`ParseFixedError`].
3. `serde-str`, disabled by default. Fixed-point numbers are serialized as
strings showing the value when using human-readable formats. This feature
requires the `serde` and the `std` optional features. **Warning:** numbers
serialized when this feature is enabled cannot be deserialized when this
feature is disabled, and vice versa.
To enable features, you can add the dependency like this to
[*Cargo.toml*]:
To enable features, you can add the dependency like this to [*Cargo.toml*]:
```toml
[dependencies.fixed]
@ -207,33 +197,29 @@ features = ["serde"]
## Experimental optional features
It is not considered a breaking change if the following experimental
features are removed. The removal of experimental features would
however require a minor version bump. Similarly, on a minor version
bump, optional dependencies can be updated to an incompatible newer
version.
It is not considered a breaking change if the following experimental features
are removed. The removal of experimental features would however require a minor
version bump. Similarly, on a minor version bump, optional dependencies can be
updated to an incompatible newer version.
1. `num-traits`, disabled by default. This implements some traits
from the [*num-traits* crate]. (The plan is to promote this to an
optional feature once the [*num-traits* crate] reaches version
1.0.0.)
1. `num-traits`, disabled by default. This implements some traits from the
[*num-traits* crate]. (The plan is to promote this to an optional feature
once the [*num-traits* crate] reaches version 1.0.0.)
## Deprecated optional features
The following optional features are deprecated and may be removed in
the next major version of the crate.
The following optional features are deprecated and may be removed in the next
major version of the crate.
1. `az`, has no effect. Previously required to enable the cast traits
provided by the [*az* crate]. Now these cast traits are always
provided.
2. `f16`, has no effect. Previously required to provide conversion
to/from [`f16`] and [`bf16`]. Now these conversions are always
provided.
1. `az`, has no effect. Previously required to enable the cast traits provided
by the [*az* crate]. Now these cast traits are always provided.
2. `f16`, has no effect. Previously required to provide conversion to/from
[`f16`] and [`bf16`]. Now these conversions are always provided.
## License
This crate is free software: you can redistribute it and/or modify it
under the terms of either
This crate is free software: you can redistribute it and/or modify it under the
terms of either
* the [Apache License, Version 2.0][LICENSE-APACHE] or
* the [MIT License][LICENSE-MIT]
@ -242,10 +228,9 @@ at your option.
### Contribution
Unless you explicitly state otherwise, any contribution intentionally
submitted for inclusion in the work by you, as defined in the Apache
License, Version 2.0, shall be dual licensed as above, without any
additional terms or conditions.
Unless you explicitly state otherwise, any contribution intentionally submitted
for inclusion in the work by you, as defined in the Apache License, Version 2.0,
shall be dual licensed as above, without any additional terms or conditions.
[*Cargo.toml*]: https://doc.rust-lang.org/cargo/guide/dependencies.html
[*az* crate]: https://crates.io/crates/az
@ -427,9 +412,9 @@ macro_rules! fixed {
if_signed_unsigned!($Signedness, "signed", "unsigned"),
" number with `Frac` fractional bits.
The number has ", $s_nbits, " bits, of which <i>f</i> = `Frac` are
fractional bits and ", $s_nbits, "  <i>f</i> are integer bits. The
value <i>x</i> can lie in the range ",
The number has ", $s_nbits, " bits, of which <i>f</i> = `Frac` are fractional
bits and ", $s_nbits, "  <i>f</i> are integer bits. The value <i>x</i> can lie
in the range ",
if_signed_unsigned!(
$Signedness,
concat!("2<sup>", $s_nbits_m1, "</sup>/2<sup><i>f</i></sup>"),
@ -455,9 +440,9 @@ When <i>f</i> = 0, <i>Δ</i> = 1 and the fixed-point number behaves like ",
if_signed_unsigned!($Signedness, "1/2  <i>x</i> < 1/2", "0  <i>x</i> < 1"),
".
`Frac` is an [`Unsigned`] as provided by the [*typenum* crate]; the
plan is to to have a major version 2 with [const generics] instead
when the Rust compiler support for them is powerful enough.
`Frac` is an [`Unsigned`] as provided by the [*typenum* crate]; the plan is to
to have a major version 2 with [const generics] instead when the Rust compiler
support for them is powerful enough.
# Examples
@ -627,14 +612,14 @@ fixed! {
FixedI128, i128, "128", False
}
/// The bit representation of a *binary128* floating-point number
/// (`f128`).
/// The bit representation of a *binary128* floating-point number (`f128`).
///
/// This type can be used to
/// * convert between fixed-point numbers and the bit representation
/// of 128-bit floating-point numbers.
/// * compare fixed-point numbers and the bit representation of
///
/// * convert between fixed-point numbers and the bit representation of
/// 128-bit floating-point numbers.
/// * compare fixed-point numbers and the bit representation of 128-bit
/// floating-point numbers.
///
/// # Examples
///
@ -667,19 +652,18 @@ impl F128Bits {
/// Defines constant fixed-point numbers from integer expressions.
///
/// This macro is useful because [`from_num`] cannot be used in
/// constant expressions.
/// This macro is useful because [`from_num`] cannot be used in constant
/// expressions.
///
/// # Alternative
///
/// The [*fixed-macro* crate] provides a convenient macro to write
/// down fixed-point constants literally in code which has two
/// advantages over this macro:
/// The [*fixed-macro* crate] provides a convenient macro to write down
/// fixed-point constants literally in code which has two advantages over this
/// macro:
///
/// 1. It can handle fixed-point numbers with fractions, not just
/// integers.
/// 2. It can be used anywhere an expression or constant expression
/// can be used, not just to define a constant.
/// 1. It can handle fixed-point numbers with fractions, not just integers.
/// 2. It can be used anywhere an expression or constant expression can be
/// used, not just to define a constant.
///
/// # Examples
///
@ -696,8 +680,7 @@ impl F128Bits {
/// ```
///
/// The following would fail to compile because
/// <code>[i32]::[MAX][i32::MAX]</code> is not representable by
/// [`I16F16`].
/// <code>[i32]::[MAX][i32::MAX]</code> is not representable by [`I16F16`].
///
/// ```rust,compile_fail
/// use fixed::{const_fixed_from_int, types::I16F16};
@ -707,10 +690,9 @@ impl F128Bits {
/// }
/// ```
///
/// The following would fail to compile because [`I16F16`] is an alias
/// for <code>[FixedI32]&lt;[U32]&gt;</code>, and this macro
/// can define [`FixedI32`] constants using [`i32`] expressions, not
/// [`i16`] expressions.
/// The following would fail to compile because [`I16F16`] is an alias for
/// <code>[FixedI32]\<[U32]></code>, and this macro can define [`FixedI32`]
/// constants using [`i32`] expressions, not [`i16`] expressions.
///
/// ```rust,compile_fail
/// use fixed::{const_fixed_from_int, types::I16F16};
@ -738,8 +720,8 @@ macro_rules! const_fixed_from_int {
)* };
}
/// These are doc tests that should not appear in the docs, but are
/// useful as doc tests can check to ensure compilation failure.
/// These are doc tests that should not appear in the docs, but are useful as
/// doc tests can check to ensure compilation failure.
///
/// The first snippet succeeds, and acts as a control.
///