improve float to_degrees/to_radians rounding comments and impl #145625

karolzwolak · 2025-08-19T20:19:37Z

This PR makes to_degrees() and to_radians() float functions more consistent between each other and improves comments around their precision and rounding.

revise comments explaining why we are using literal or expression
add unspecified precision comments as we don't guarantee precision
use expression in f128::to_degrees()
make f64::to_degrees() impl consistent with other functions

r? @tgross35

tgross35 · 2025-08-19T20:23:17Z

library/core/src/num/f16.rs

    pub const fn to_degrees(self) -> Self {
-        // Use a literal for better precision.
-        const PIS_IN_180: f16 = 57.2957795130823208767981548141051703_f16;
-        self * PIS_IN_180
+        self * 57.2957795130823208767981548141051703_f16
    }


The results here will be identical. I meant something like what f64 does where the result is calculated (possibly to a const),

rust/library/core/src/num/f64.rs

Lines 871 to 876 in 16ad385

pub const fn to_degrees(self) -> f64 {

// The division here is correctly rounded with respect to the true

// value of 180/π. (This differs from f32, where a constant must be

// used to ensure a correctly rounded result.)

self * (180.0f64 / consts::PI)

}

So you're saying we can just merge this? And why the behavior difference for f64?

Oh wait, you mean the to_degrees() is a const fn, which in turn could increase precision for constant which uses to_degrees()?

Not quite :) I mean that it makes absolutely no difference whether we multiply by the literal 57.2957795130823208767981548141051703, or assign it to a const first, or let. They compiler guarantees that the comptime and runtime semantics are the same here.

What I do mean is for f16, 57.295... is not exactly the same as 180 / pi https://play.rust-lang.org/?version=nightly&mode=debug&edition=2024&gist=85a373861e735f8c2a58b00b4540a150. And using the value of 180 / pi happens to roundtrip for an input of 180.0, though not for other values.

I'm not sure which value (exact at larger precision or computed as f16) should be preferred though, @beetrees is more likely to have this knowledge ready to go.

The exact value of 180 / pi (57.295...) should be used. This should give more accurate results most of the time as rounding occurs fewer times.

Not directly related to this PR: the to_degrees and to_radians functions on floats don't currently document the precision of the result. They should probably have an "Unspecified precision" note similar to most other mathematical functions, as we don't currently provide any guarantees.

Then we should use the exact value for f32 as well right?

rust/library/core/src/num/f32.rs

Lines 874 to 877 in ed06877

pub const fn to_radians(self) -> f32 {

const RADS_PER_DEG: f32 = consts::PI / 180.0;

self * RADS_PER_DEG

}

And why is it okay for f64?

rust/library/core/src/num/f64.rs

Lines 871 to 876 in ed06877

pub const fn to_degrees(self) -> f64 {

// The division here is correctly rounded with respect to the true

// value of 180/π. (This differs from f32, where a constant must be

// used to ensure a correctly rounded result.)

self * (180.0f64 / consts::PI)

}

Is this simply because f64 has enough precision?
If so then why aren't we doing the same for f128?

rust/library/core/src/num/f128.rs

Lines 647 to 651 in ed06877

pub const fn to_degrees(self) -> Self {

// Use a literal for better precision.

const PIS_IN_180: f128 = 57.2957795130823208767981548141051703324054724665643215491602_f128;

self * PIS_IN_180

}

Okay, so I think we should:

leave the constant literals

replace the comments

// Use a literal for better precision.

with something along the lines

// Use a literal to avoid division rounding

add Unspecified precision comment to to_degrees and to_radians conversions for all floats

Is this simply because f64 has enough precision?
If so then why aren't we doing the same for f128?

I've tested it, and 180.0 / consts::PI == 57.2957795130823208767981548141051703 is false for f16 and f32 but true for f64 and f128. Therefore f128::to_degrees can also just use 180.0 / consts::PI like f64::to_degrees does.

with something along the lines

The comment should probably mention that using the literal avoids the double rounding that occurs when using 180.0 / consts::PI to calculate the constant (the value of consts::PI is already rounded from the true irrational value of π, with the division then rounding again as you've stated. For f64 and f128, it just coincidentally happens that the double-rounded value of 180.0 / consts::PI is the same as the correctly rounded exact value).

I also tested pi / 180,

f16: false f32: true f64: true f128: false

and it matches with the to_radians implementation, f32 and f64 use division.

rustbot · 2025-08-20T19:40:27Z

Requested reviewer is already assigned to this pull request.

Please choose another assignee.

karolzwolak · 2025-08-20T19:43:11Z

I tried to address all of changes we discussed.

rustbot · 2025-08-20T19:43:13Z

Failed to set assignee to beetrees: invalid assignee

Note: Only org members with at least the repository "read" role, users with write permissions, or people who have commented on the PR may be assigned.

beetrees

Thanks

library/core/src/num/f128.rs

karolzwolak · 2025-08-21T15:22:51Z

update:

don't mention the other float types
add examples headings

beetrees

LGTM

View changes since this review

tgross35

Thanks @beetrees for reviewing.

One remaining nit @karolzwolak, could you rewrap comments to 100 chars? Both doc and normal comments seem to be ~115.

View changes since this review

tgross35 · 2025-08-21T21:10:11Z

Please squash as well

* revise comments explaining why we are using literal or expression * add unspecified precision comments as we don't guarantee precision * use expression in `f128::to_degrees()` * make `f64::to_degrees()` impl consistent with other functions

karolzwolak · 2025-08-22T13:49:03Z

Done. @rustbot review

tgross35 · 2025-08-27T08:29:05Z

Looks great, thank you!

@bors r=beetrees,tgross35

bors · 2025-08-27T08:29:08Z

📌 Commit 698db13 has been approved by beetrees,tgross35

It is now in the queue for this repository.

Rollup of 6 pull requests Successful merges: - #144274 (add Option::reduce) - #145562 (Simplify macro generating ToString implementations for `&…&str`) - #145625 (improve float to_degrees/to_radians rounding comments and impl) - #145740 (Introduce a `[workspace.dependencies`] section in the top-level `Cargo.toml`) - #145885 (Inherit TCC in debuginfo tests on macOS) - #145905 (Stop calling unwrap when format foreign has trailing dollar) r? `@ghost` `@rustbot` modify labels: rollup

Rollup merge of #145625 - karolzwolak:f16-use-expr-instead-literal, r=beetrees,tgross35 improve float to_degrees/to_radians rounding comments and impl This PR makes `to_degrees()` and `to_radians()` float functions more consistent between each other and improves comments around their precision and rounding. * revise comments explaining why we are using literal or expression * add unspecified precision comments as we don't guarantee precision * use expression in `f128::to_degrees()` * make `f64::to_degrees()` impl consistent with other functions r? `@tgross35`

rustbot assigned tgross35 Aug 19, 2025

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Aug 19, 2025

tgross35 reviewed Aug 19, 2025

View reviewed changes

karolzwolak force-pushed the f16-use-expr-instead-literal branch from ed06877 to f734a50 Compare August 20, 2025 19:38

karolzwolak changed the title ~~simplify f16 to_degrees and to_radians by using expression directly~~ improve float to_degrees/to_radians rounding comments and impl Aug 20, 2025

beetrees suggested changes Aug 20, 2025

View reviewed changes

library/core/src/num/f128.rs Show resolved Hide resolved

library/core/src/num/f128.rs Outdated Show resolved Hide resolved

karolzwolak force-pushed the f16-use-expr-instead-literal branch from f734a50 to 21a7f52 Compare August 21, 2025 15:22

tgross35 requested a review from beetrees August 21, 2025 19:35

beetrees approved these changes Aug 21, 2025

View reviewed changes

tgross35 requested changes Aug 21, 2025

View reviewed changes

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Aug 21, 2025

karolzwolak force-pushed the f16-use-expr-instead-literal branch from 21a7f52 to 698db13 Compare August 22, 2025 13:47

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Aug 22, 2025

tgross35 approved these changes Aug 27, 2025

View reviewed changes

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Aug 27, 2025

matthiaskrgr mentioned this pull request Aug 27, 2025

Rollup of 6 pull requests #145923

Merged

bors merged commit c0cd29e into rust-lang:master Aug 27, 2025
10 checks passed

rustbot added this to the 1.91.0 milestone Aug 27, 2025

	pub const fn to_degrees(self) -> f64 {
	// The division here is correctly rounded with respect to the true
	// value of 180/π. (This differs from f32, where a constant must be
	// used to ensure a correctly rounded result.)
	self * (180.0f64 / consts::PI)
	}

	pub const fn to_radians(self) -> f32 {
	const RADS_PER_DEG: f32 = consts::PI / 180.0;
	self * RADS_PER_DEG
	}

	pub const fn to_degrees(self) -> Self {
	// Use a literal for better precision.
	const PIS_IN_180: f128 = 57.2957795130823208767981548141051703324054724665643215491602_f128;
	self * PIS_IN_180
	}

improve float to_degrees/to_radians rounding comments and impl #145625

improve float to_degrees/to_radians rounding comments and impl #145625

Conversation

karolzwolak commented Aug 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tgross35 Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

karolzwolak Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

karolzwolak Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

tgross35 Aug 19, 2025

Choose a reason for hiding this comment

Uh oh!

beetrees Aug 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

karolzwolak Aug 20, 2025

Choose a reason for hiding this comment

Uh oh!

karolzwolak Aug 20, 2025

Choose a reason for hiding this comment

Uh oh!

beetrees Aug 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

karolzwolak Aug 20, 2025

Choose a reason for hiding this comment

Uh oh!

rustbot commented Aug 20, 2025

Uh oh!

karolzwolak commented Aug 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Aug 20, 2025

Uh oh!

beetrees left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

karolzwolak commented Aug 21, 2025

Uh oh!

beetrees left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tgross35 left a comment • edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tgross35 commented Aug 21, 2025

Uh oh!

karolzwolak commented Aug 22, 2025

Uh oh!

tgross35 commented Aug 27, 2025

Uh oh!

bors commented Aug 27, 2025

Uh oh!

Uh oh!

Uh oh!

karolzwolak commented Aug 19, 2025 •

edited

Loading

beetrees Aug 19, 2025 •

edited

Loading

beetrees Aug 20, 2025 •

edited

Loading

karolzwolak commented Aug 20, 2025 •

edited

Loading

beetrees left a comment •

edited by rustbot

Loading

tgross35 left a comment •

edited by rustbot

Loading