Skip to content

refactor(xor): remove persisted item counts - #795

Merged
tisonkun merged 4 commits into
apache:mainfrom
tisonkun:codex/xor-remove-item-count
Nov 8, 2026
Merged

tisonkun merged 4 commits into
apache:mainfrom
tisonkun:codex/xor-remove-item-count

Conversation

@tisonkun

@tisonkun tisonkun commented Nov 5, 2026 •

Copy link
Copy Markdown
Member

XOR membership queries need the construction seed, segment layout, and fingerprints, but not the number of input hashes. Remove the finished filter's count and getNumItems(), isEmpty(), and getBitsPerItem(). The builder still tracks its updates.

The serialized header occupies 24 bytes: flags at byte 3, fingerprint width at byte 4, seed at byte 8, and segment length at byte 16. Remaining bytes are reserved. Writers emit zero flags; readers reject unsupported flags. The layout has three fingerprint segments and no persisted item count.

Release compatibility needs resolution before merging. Java 9.1.0-RC1 contains the count-bearing version-1 format and the three public methods. This PR currently retains serialization version 1 and removes those methods. If that release ships first, this PR needs a distinct format version with a legacy reader, and an agreed API migration. It must not be released as a compatible minor update in its current form.

Validation: mvn -B clean verify passes all 2,335 tests. Coverage includes heapify/wrap round trips, reference images, unsupported flags, truncated images, fingerprint widths, hash count, and capacity invariants. The preamble error message now refers to longs rather than bytes.

@tisonkun

tisonkun commented Nov 5, 2026 •

Copy link
Copy Markdown
Member Author

The Rust counterpart uses the same header and serialization version 1. Its serialization tests consume scalar and UTF-8/raw-byte fixtures generated from this PR's c179c5bc0 and verify byte-for-byte construction and round trips.

@leerho

leerho commented Nov 8, 2026

Copy link
Copy Markdown
Member

We have just found a critical fix. So this RC-1. is cancelled.

@PsiACE PsiACE left a comment

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lgtm

@tisonkun
tisonkun merged commit ef440a8 into apache:main Nov 8, 2026
6 checks passed
@tisonkun
tisonkun deleted the codex/xor-remove-item-count branch November 8, 2026 06:19
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants