Struct BlockRef 

pub struct BlockRef {
    pub dir: PathBuf,
    pub min_ts: i64,
    pub max_ts: i64,
    pub node: u32,
    pub seq: u64,
    pub wal_hi: u64,
}

A published block, discovered by reading directory names.

Fields

dir: PathBuf

min_ts: i64

max_ts: i64

node: u32

Which replica wrote this block. See node_id.

seq: u64

wal_hi: u64

One past the highest write-ahead-log sequence whose records are in this block, or 0 for a block published before the log existed.

This is the entire manifest. Principle 4 says there is no coordination state and the block directory is the manifest, so the one fact WAL replay needs — how far the log has been absorbed — is carried in the name rather than in a file someone has to keep consistent with the blocks. Recovery therefore stays the readdir section 4 promises it is.

Implementations

impl BlockRef

pub fn overlaps(&self, from: i64, to: i64) -> bool

True if this block could contain a row in [from, to]. The whole point of the naming scheme: pruning without opening a single file.

Trait Implementations

impl Clone for BlockRef

fn clone(&self) -> BlockRef

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for BlockRef

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

Formats the value using the given formatter.

impl Eq for BlockRef

impl PartialEq for BlockRef

fn eq(&self, other: &BlockRef) -> 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 StructuralPartialEq for BlockRef