Improve doc formatting and linking.

src/angle.rs

@@ -132,7 +132,7 @@ impl<T> Angle<T>
 where
     T: Float,
 {
-    /// Returns true if the angle is a finite number.
+    /// Returns `true` if the angle is a finite number.
     #[inline]
     pub fn is_finite(self) -> bool {
         self.radians.is_finite()
@@ -143,7 +143,7 @@ impl<T> Angle<T>
 where
     T: Real,
 {
-    /// Returns (sin(self), cos(self)).
+    /// Returns `(sin(self), cos(self))`.
     pub fn sin_cos(self) -> (T, T) {
         self.radians.sin_cos()
     }

src/approxeq.rs

@@ -12,12 +12,12 @@ pub trait ApproxEq<Eps> {
     /// Default epsilon value
     fn approx_epsilon() -> Eps;
 
-    /// Returns `true` is this object is approximately equal to the other one, using
+    /// Returns `true` if this object is approximately equal to the other one, using
     /// a provided epsilon value.
     fn approx_eq_eps(&self, other: &Self, approx_epsilon: &Eps) -> bool;
 
-    /// Returns `true` is this object is approximately equal to the other one, using
-    /// the `approx_epsilon()` epsilon value.
+    /// Returns `true` if this object is approximately equal to the other one, using
+    /// the [`approx_epsilon`](ApproxEq::approx_epsilon) epsilon value.
     fn approx_eq(&self, other: &Self) -> bool {
         self.approx_eq_eps(other, &Self::approx_epsilon())
     }

src/box2d.rs

@@ -53,11 +53,10 @@ use core::ops::{Add, Div, DivAssign, Mul, MulAssign, Range, Sub};
 /// - it's area is negative (`min.x > max.x` or `min.y > max.y`),
 /// - it contains NaNs.
 ///
-/// [`Rect`]: struct.Rect.html
-/// [`intersection`]: #method.intersection
-/// [`is_empty`]: #method.is_empty
-/// [`union`]: #method.union
-/// [`size`]: #method.size
+/// [`intersection`]: Self::intersection
+/// [`is_empty`]: Self::is_empty
+/// [`union`]: Self::union
+/// [`size`]: Self::size
 #[repr(C)]
 #[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
 #[cfg_attr(
@@ -139,7 +138,7 @@ impl<T, U> Box2D<T, U> {
         }
     }
 
-    /// Creates a Box2D of the given size, at offset zero.
+    /// Creates a `Box2D` of the given size, at offset zero.
     #[inline]
     pub fn from_size(size: Size2D<T, U>) -> Self
     where
@@ -156,7 +155,7 @@ impl<T, U> Box2D<T, U>
 where
     T: PartialOrd,
 {
-    /// Returns true if the box has a negative area.
+    /// Returns `true` if the box has a negative area.
     ///
     /// The common interpretation for a negative box is to consider it empty. It can be obtained
     /// by calculating the intersection of two boxes that do not intersect.
@@ -165,7 +164,7 @@ where
         self.max.x < self.min.x || self.max.y < self.min.y
     }
 
-    /// Returns true if the size is zero, negative or NaN.
+    /// Returns `true` if the size is zero, negative or NaN.
     #[inline]
     pub fn is_empty(&self) -> bool {
         !(self.max.x > self.min.x && self.max.y > self.min.y)
@@ -240,7 +239,7 @@ where
     ///
     /// The result is a negative box if the boxes do not intersect.
     /// This can be useful for computing the intersection of more than two boxes, as
-    /// it is possible to chain multiple intersection_unchecked calls and check for
+    /// it is possible to chain multiple `intersection_unchecked` calls and check for
     /// empty/negative result at the end.
     #[inline]
     pub fn intersection_unchecked(&self, other: &Self) -> Self {
@@ -553,7 +552,11 @@ impl<T: NumCast + Copy, U> Box2D<T, U> {
     ///
     /// When casting from floating point to integer coordinates, the decimals are truncated
     /// as one would expect from a simple cast, but this behavior does not always make sense
-    /// geometrically. Consider using round(), round_in or round_out() before casting.
+    /// geometrically. Consider using [`round`], [`round_in`] or [`round_out`] before casting.
+    ///
+    /// [`round`]: Self::round
+    /// [`round_in`]: Self::round_in
+    /// [`round_out`]: Self::round_out
     #[inline]
     pub fn cast<NewT: NumCast>(&self) -> Box2D<NewT, U> {
         Box2D::new(self.min.cast(), self.max.cast())
@@ -563,7 +566,11 @@ impl<T: NumCast + Copy, U> Box2D<T, U> {
     ///
     /// When casting from floating point to integer coordinates, the decimals are truncated
     /// as one would expect from a simple cast, but this behavior does not always make sense
-    /// geometrically. Consider using round(), round_in or round_out() before casting.
+    /// geometrically. Consider using [`round`], [`round_in`] or [`round_out`] before casting.
+    ///
+    /// [`round`]: Self::round
+    /// [`round_in`]: Self::round_in
+    /// [`round_out`]: Self::round_out
     pub fn try_cast<NewT: NumCast>(&self) -> Option<Box2D<NewT, U>> {
         match (self.min.try_cast(), self.max.try_cast()) {
             (Some(a), Some(b)) => Some(Box2D::new(a, b)),
@@ -627,7 +634,7 @@ impl<T: NumCast + Copy, U> Box2D<T, U> {
 }
 
 impl<T: Float, U> Box2D<T, U> {
-    /// Returns true if all members are finite.
+    /// Returns `true` if all members are finite.
     #[inline]
     pub fn is_finite(self) -> bool {
         self.min.is_finite() && self.max.is_finite()

src/box3d.rs

@@ -113,7 +113,7 @@ impl<T, U> Box3D<T, U> {
         }
     }
 
-    /// Creates a Box3D of the given size, at offset zero.
+    /// Creates a `Box3D` of the given size, at offset zero.
     #[inline]
     pub fn from_size(size: Size3D<T, U>) -> Self
     where
@@ -130,7 +130,7 @@ impl<T, U> Box3D<T, U>
 where
     T: PartialOrd,
 {
-    /// Returns true if the box has a negative volume.
+    /// Returns `true` if the box has a negative volume.
     ///
     /// The common interpretation for a negative box is to consider it empty. It can be obtained
     /// by calculating the intersection of two boxes that do not intersect.
@@ -139,7 +139,7 @@ where
         self.max.x < self.min.x || self.max.y < self.min.y || self.max.z < self.min.z
     }
 
-    /// Returns true if the size is zero, negative or NaN.
+    /// Returns `true` if the size is zero, negative or NaN.
     #[inline]
     pub fn is_empty(&self) -> bool {
         !(self.max.x > self.min.x && self.max.y > self.min.y && self.max.z > self.min.z)
@@ -556,7 +556,11 @@ impl<T: NumCast + Copy, U> Box3D<T, U> {
     ///
     /// When casting from floating point to integer coordinates, the decimals are truncated
     /// as one would expect from a simple cast, but this behavior does not always make sense
-    /// geometrically. Consider using round(), round_in or round_out() before casting.
+    /// geometrically. Consider using [`round`], [`round_in`] or [`round_out`] before casting.
+    ///
+    /// [`round`]: Self::round
+    /// [`round_in`]: Self::round_in
+    /// [`round_out`]: Self::round_out
     #[inline]
     pub fn cast<NewT: NumCast>(&self) -> Box3D<NewT, U> {
         Box3D::new(self.min.cast(), self.max.cast())
@@ -566,7 +570,11 @@ impl<T: NumCast + Copy, U> Box3D<T, U> {
     ///
     /// When casting from floating point to integer coordinates, the decimals are truncated
     /// as one would expect from a simple cast, but this behavior does not always make sense
-    /// geometrically. Consider using round(), round_in or round_out() before casting.
+    /// geometrically. Consider using [`round`], [`round_in`] or [`round_out`] before casting.
+    ///
+    /// [`round`]: Self::round
+    /// [`round_in`]: Self::round_in
+    /// [`round_out`]: Self::round_out
     pub fn try_cast<NewT: NumCast>(&self) -> Option<Box3D<NewT, U>> {
         match (self.min.try_cast(), self.max.try_cast()) {
             (Some(a), Some(b)) => Some(Box3D::new(a, b)),
@@ -630,7 +638,7 @@ impl<T: NumCast + Copy, U> Box3D<T, U> {
 }
 
 impl<T: Float, U> Box3D<T, U> {
-    /// Returns true if all members are finite.
+    /// Returns `true` if all members are finite.
     #[inline]
     pub fn is_finite(self) -> bool {
         self.min.is_finite() && self.max.is_finite()

src/homogen.rs

@@ -143,7 +143,7 @@ impl<T, U> HomogeneousVector<T, U> {
 impl<T: Copy + Div<T, Output = T> + Zero + PartialOrd, U> HomogeneousVector<T, U> {
     /// Convert into Cartesian 2D point.
     ///
-    /// Returns None if the point is on or behind the W=0 hemisphere.
+    /// Returns `None` if the point is on or behind the W=0 hemisphere.
     #[inline]
     pub fn to_point2d(self) -> Option<Point2D<T, U>> {
         if self.w > T::zero() {
@@ -155,7 +155,7 @@ impl<T: Copy + Div<T, Output = T> + Zero + PartialOrd, U> HomogeneousVector<T, U
 
     /// Convert into Cartesian 3D point.
     ///
-    /// Returns None if the point is on or behind the W=0 hemisphere.
+    /// Returns `None` if the point is on or behind the W=0 hemisphere.
     #[inline]
     pub fn to_point3d(self) -> Option<Point3D<T, U>> {
         if self.w > T::zero() {

src/length.rs

@@ -36,10 +36,8 @@ use serde::{Deserialize, Deserializer, Serialize, Serializer};
 /// expression that requires a different unit.  It may be a type without values, such as an empty
 /// enum.
 ///
-/// You can multiply a `Length` by a `scale::Scale` to convert it from one unit to
+/// You can multiply a `Length` by a [`Scale`] to convert it from one unit to
 /// another. See the [`Scale`] docs for an example.
-///
-/// [`Scale`]: struct.Scale.html
 #[repr(C)]
 pub struct Length<T, Unit>(pub T, #[doc(hidden)] pub PhantomData<Unit>);
 

src/point.rs

@@ -148,7 +148,7 @@ impl<T, U> Point2D<T, U> {
         point2(Zero::zero(), Zero::zero())
     }
 
-    /// The same as [`origin()`](#method.origin).
+    /// The same as [`Point2D::origin`].
     #[inline]
     pub fn zero() -> Self
     where
@@ -509,7 +509,7 @@ impl<T: NumCast + Copy, U> Point2D<T, U> {
         self.cast()
     }
 
-    /// Cast into an i32 point, truncating decimals if any.
+    /// Cast into an `i32` point, truncating decimals if any.
     ///
     /// When casting from floating point points, it is worth considering whether
     /// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
@@ -519,7 +519,7 @@ impl<T: NumCast + Copy, U> Point2D<T, U> {
         self.cast()
     }
 
-    /// Cast into an i64 point, truncating decimals if any.
+    /// Cast into an `i64` point, truncating decimals if any.
     ///
     /// When casting from floating point points, it is worth considering whether
     /// to `round()`, `ceil()` or `floor()` before the cast in order to obtain
@@ -531,7 +531,7 @@ impl<T: NumCast + Copy, U> Point2D<T, U> {
 }
 
 impl<T: Float, U> Point2D<T, U> {
-    /// Returns true if all members are finite.
+    /// Returns `true` if all members are finite.
     #[inline]
     pub fn is_finite(self) -> bool {
         self.x.is_finite() && self.y.is_finite()
@@ -710,23 +710,23 @@ impl<T: Zero, U> Zero for Point2D<T, U> {
 }
 
 impl<T: Round, U> Round for Point2D<T, U> {
-    /// See [Point2D::round()](#method.round)
+    /// See [`Point2D::round`].
     #[inline]
     fn round(self) -> Self {
         self.round()
     }
 }
 
 impl<T: Ceil, U> Ceil for Point2D<T, U> {
-    /// See [Point2D::ceil()](#method.ceil)
+    /// See [`Point2D::ceil`].
     #[inline]
     fn ceil(self) -> Self {
         self.ceil()
     }
 }
 
 impl<T: Floor, U> Floor for Point2D<T, U> {
-    /// See [Point2D::floor()](#method.floor)
+    /// See [`Point2D::floor`].
     #[inline]
     fn floor(self) -> Self {
         self.floor()
@@ -944,7 +944,7 @@ impl<T, U> Point3D<T, U> {
         point3(Zero::zero(), Zero::zero(), Zero::zero())
     }
 
-    /// The same as [`origin()`](#method.origin).
+    /// The same as [`Point3D::origin`].
     #[inline]
     pub fn zero() -> Self
     where
@@ -1356,7 +1356,7 @@ impl<T: NumCast + Copy, U> Point3D<T, U> {
 }
 
 impl<T: Float, U> Point3D<T, U> {
-    /// Returns true if all members are finite.
+    /// Returns `true` if all members are finite.
     #[inline]
     pub fn is_finite(self) -> bool {
         self.x.is_finite() && self.y.is_finite() && self.z.is_finite()
@@ -1551,23 +1551,23 @@ impl<T: Zero, U> Zero for Point3D<T, U> {
 }
 
 impl<T: Round, U> Round for Point3D<T, U> {
-    /// See [Point3D::round()](#method.round)
+    /// See [`Point3D::round`].
     #[inline]
     fn round(self) -> Self {
         self.round()
     }
 }
 
 impl<T: Ceil, U> Ceil for Point3D<T, U> {
-    /// See [Point3D::ceil()](#method.ceil)
+    /// See [`Point3D::ceil`].
     #[inline]
     fn ceil(self) -> Self {
         self.ceil()
     }
 }
 
 impl<T: Floor, U> Floor for Point3D<T, U> {
-    /// See [Point3D::floor()](#method.floor)
+    /// See [`Point3D::floor`].
     #[inline]
     fn floor(self) -> Self {
         self.floor()