Struct selectors::parser::AncestorHashes

pub struct AncestorHashes {
    pub packed_hashes: [u32; 3],
}

Ancestor hashes for the bloom filter. We precompute these and store them inline with selectors to optimize cache performance during matching. This matters a lot.

We use 4 hashes, which is copied from Gecko, who copied it from WebKit. Note that increasing the number of hashes here will adversely affect the cache hit when fast-rejecting long lists of Rules with inline hashes.

Because the bloom filter only uses the bottom 24 bits of the hash, we pack the fourth hash into the upper bits of the first three hashes in order to shrink Rule (whose size matters a lot). This scheme minimizes the runtime overhead of the packing for the first three hashes (we just need to mask off the upper bits) at the expense of making the fourth somewhat more complicated to assemble, because we often bail out before checking all the hashes.

Fields

packed_hashes: [u32; 3]

Implementations

impl AncestorHashes

pub fn new<Impl: SelectorImpl>( selector: &Selector<Impl>, quirks_mode: QuirksMode, ) -> Self

pub fn fourth_hash(&self) -> u32

Returns the fourth hash, reassembled from parts.

Trait Implementations

impl Clone for AncestorHashes

fn clone(&self) -> AncestorHashes

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for AncestorHashes

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for AncestorHashes

fn eq(&self, other: &AncestorHashes) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Eq for AncestorHashes

impl StructuralPartialEq for AncestorHashes