pub enum Focus<'a, A> {
// some variants omitted
}
Expand description
Focused indexing over a Vector
.
By remembering the last tree node accessed through an index lookup and the path we took to get there, we can speed up lookups for adjacent indices tremendously. Lookups on indices in the same node are instantaneous, and lookups on sibling nodes are also very fast.
A Focus
can also be used as a restricted view into a vector, using the
narrow
and split_at
methods.
When should I use a Focus
for better performance?
Focus
is useful when you need to perform a large number of index lookups
that are more likely than not to be close to each other. It’s usually worth
using a Focus
in any situation where you’re batching a lot of index
lookups together, even if they’re not obviously adjacent - there’s likely
to be some performance gain for even completely random access.
If you’re just iterating forwards or backwards over the Vector
in order, you’re better off with a regular iterator, which, in fact, is
implemented using a Focus
, but provides a simpler interface.
If you’re just doing a very small number of index lookups, the setup cost
for the Focus
is probably not worth it.
A Focus
is never faster than an index lookup on a small Vector
with a length below the internal RRB tree’s branching factor of 64.
Examples
This example is contrived, as the better way to iterate forwards or
backwards over a vector is with an actual iterator. Even so, the version
using a Focus
should run nearly an order of magnitude faster than the
version using index lookups at a length of 1000. It should also be noted
that vector::Iter
is actually implemented using a Focus
behind
the scenes, so the performance of the two should be identical.
let mut vec: Vector<i64> = Vector::from_iter(0..1000);
// Summing a vector, the slow way:
let mut sum = 0;
for i in 0..1000 {
sum += *vec.get(i).unwrap();
}
assert_eq!(499500, sum);
// Summing a vector faster using a Focus:
let mut sum = 0;
let mut focus = vec.focus();
for i in 0..1000 {
sum += *focus.get(i).unwrap();
}
assert_eq!(499500, sum);
// And the easy way, for completeness:
let sum: i64 = vec.iter().sum();
assert_eq!(499500, sum);
Implementations
sourceimpl<'a, A> Focus<'a, A> where
A: Clone + 'a,
impl<'a, A> Focus<'a, A> where
A: Clone + 'a,
sourcepub fn get(&mut self, index: usize) -> Option<&A>
pub fn get(&mut self, index: usize) -> Option<&A>
Get a reference to the value at a given index.
sourcepub fn index(&mut self, index: usize) -> &A
pub fn index(&mut self, index: usize) -> &A
Get a reference to the value at a given index.
Panics if the index is out of bounds.
sourcepub fn chunk_at(&mut self, index: usize) -> (Range<usize>, &[A])
pub fn chunk_at(&mut self, index: usize) -> (Range<usize>, &[A])
Get the chunk for the given index.
This gives you a reference to the leaf node that contains the index, along with its start and end indices.
sourcepub fn narrow<R>(self, range: R) -> Self where
R: RangeBounds<usize>,
pub fn narrow<R>(self, range: R) -> Self where
R: RangeBounds<usize>,
Narrow the focus onto a subslice of the vector.
Focus::narrow(range)
has the same effect as &slice[range]
, without
actually modifying the underlying vector.
Panics if the range isn’t fully inside the current focus.
Examples
let vec = Vector::from_iter(0..1000);
let narrowed = vec.focus().narrow(100..200);
let narrowed_vec = narrowed.into_iter().cloned().collect();
assert_eq!(Vector::from_iter(100..200), narrowed_vec);
sourcepub fn split_at(self, index: usize) -> (Self, Self)
pub fn split_at(self, index: usize) -> (Self, Self)
Split the focus into two.
Given an index index
, consume the focus and produce two new foci, the
left onto indices 0..index
, and the right onto indices index..N
where N
is the length of the current focus.
Panics if the index is out of bounds.
This is the moral equivalent of slice::split_at
, in
that it leaves the underlying data structure unchanged, unlike
Vector::split_at
.
Examples
let vec = Vector::from_iter(0..1000);
let (left, right) = vec.focus().split_at(500);
let left_vec = left.into_iter().cloned().collect();
let right_vec = right.into_iter().cloned().collect();
assert_eq!(Vector::from_iter(0..500), left_vec);
assert_eq!(Vector::from_iter(500..1000), right_vec);
Trait Implementations
Auto Trait Implementations
impl<'a, A> RefUnwindSafe for Focus<'a, A> where
A: RefUnwindSafe,
impl<'a, A> !Send for Focus<'a, A>
impl<'a, A> !Sync for Focus<'a, A>
impl<'a, A> Unpin for Focus<'a, A>
impl<'a, A> UnwindSafe for Focus<'a, A> where
A: RefUnwindSafe,
Blanket Implementations
sourceimpl<T> BorrowMut<T> for T where
T: ?Sized,
impl<T> BorrowMut<T> for T where
T: ?Sized,
const: unstable · sourcefn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Mutably borrows from an owned value. Read more
sourceimpl<T> ToOwned for T where
T: Clone,
impl<T> ToOwned for T where
T: Clone,
type Owned = T
type Owned = T
The resulting type after obtaining ownership.
sourcefn clone_into(&self, target: &mut T)
fn clone_into(&self, target: &mut T)
toowned_clone_into
)Uses borrowed data to replace owned data, usually by cloning. Read more