You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This follows how segments are implemented where each segment represents a current snapshot or delta of a log (composed of a set of blocks containing the facts). Segments point to the segment they are based on (also referred to as the parent) as well as the next segment to follow (which is usually the same as base).
To enable reuse across segments, blocks should be addressable independent of the segment they were originally created for. This enables new segments to reference existing blocks that have not changed between revisions. For the standard log-based index (ordered by transaction), blocks will never need to be replaced since this is the natural ordering. For sorted indexes, any block that comes before the new facts will need to be reindexed. This enables reuse of existing blocks that have not changed and requires only creating a few new blocks for subsequent segments.
More formally, this is known as a persistent data structure where copies or revisions of a value internally point to existing data rather than physically copying the data.
The high-level requirements for this are:
Add a Block struct type which defines a Next field for pointing to the next block in the sequence.
Add a Head field to the Segment structure that points to the first block in the sequence.
The text was updated successfully, but these errors were encountered:
This follows how segments are implemented where each segment represents a current snapshot or delta of a log (composed of a set of blocks containing the facts). Segments point to the segment they are based on (also referred to as the parent) as well as the next segment to follow (which is usually the same as
base
).To enable reuse across segments, blocks should be addressable independent of the segment they were originally created for. This enables new segments to reference existing blocks that have not changed between revisions. For the standard log-based index (ordered by transaction), blocks will never need to be replaced since this is the natural ordering. For sorted indexes, any block that comes before the new facts will need to be reindexed. This enables reuse of existing blocks that have not changed and requires only creating a few new blocks for subsequent segments.
More formally, this is known as a persistent data structure where copies or revisions of a value internally point to existing data rather than physically copying the data.
The high-level requirements for this are:
Block
struct type which defines aNext
field for pointing to the next block in the sequence.Head
field to theSegment
structure that points to the first block in the sequence.The text was updated successfully, but these errors were encountered: