Skip to content

Repros 744 - #748

Merged
tlwillke merged 8 commits into
batched-vector-accessfrom
repros-744
Oct 12, 2026
Merged

Repros 744#748
tlwillke merged 8 commits into
batched-vector-accessfrom
repros-744

Conversation

@jshook

@jshook jshook commented Oct 12, 2026

Copy link
Copy Markdown
Contributor

Some review feedback for #744 in the form of unit tests for reproducing issues and fixes that pair with them.

jshook added 6 commits October 12, 2026 20:27
@jshook
Add failing repro: explicit dataset_loader=buffered is downgraded to …
b99d57a 
…preload

BaseVectorLoading.forBenchmark(MultiConfig) returns PRELOAD for configs with
search or construction compression None even when -Djvector.dataset_loader=buffered
is set, although the docs, release note and PR description call the property an
explicit override. Fails at 87bb453 (PR #744).
@jshook
Add failing repro: full-precision build scoring re-reads the anchor p…
0145908 
…er comparison

For a value-shared source, randomAccessScoreProvider's score functions read both
the anchor and the compared node on every similarityTo call: 2000 source reads
for 1000 comparisons against one anchor. Fails at 87bb453 (PR #744).
@jshook
Add failing repro: closing FvecFileVectorValues under point-read load
1f9fa86 
Storage.close() clears point readers' channels and buffers without taking their
leases, so in-flight point reads fail with NullPointerException and can reopen a
channel after close. Timing-dependent; fails within a few rounds here at
87bb453 (PR #744).
@jshook
Honor an explicit jvector.dataset_loader over benchmark source routing
581befd 
forBenchmark(MultiConfig) now returns an explicitly set jvector.dataset_loader
as-is, so buffered applies even to uncompressed-construction and exact-search
configurations, matching the documented explicit override. Routing still
applies when buffered input comes from the YAML useDatasetPrefetching flag.
Fixes TestDatasetLoaderOverride.
@jshook
Read the exact-scoring anchor once per score function for shared sources
e9aecad 
randomAccessScoreProvider's value-shared path re-read node1 on every
comparison, doubling source reads in full-precision builds from a file-backed
source. The anchor is now copied on first use and kept by the score function:
creation still reads nothing, the copy can't be overwritten by later reads
through the same views, and each comparison reads only node2.
Fixes TestBuildScoreProviderAnchorReads.
@jshook
Make closing FvecFileVectorValues safe against in-flight point reads
a6a578d 
Storage.close() cleared each point reader's channel and buffer without taking
its lease, so a concurrent point read could hit a null buffer or reopen a
channel that was never closed. close() now takes each reader's lease before
closing its channel, and readPointInto checks for closure after taking a
lease, so later point reads fail with IllegalStateException.
Fixes TestFvecCloseDuringPointReads.
@github-actions

github-actions Bot commented Oct 12, 2026

Copy link
Copy Markdown
Contributor

Before you submit for review:

  • Does your PR follow guidelines from CONTRIBUTIONS.md?
  • Did you summarize what this PR does clearly and concisely?
  • Did you include performance data for changes which may be performance impacting?
  • Did you include useful docs for any user-facing changes or features?
  • Did you include useful javadocs for developer oriented changes, explaining new concepts or key changes?
  • Did you rebase your branch onto the latest main for regression testing and PR submission?
  • Did you trigger regression testing via Run Bench Main and review results?
  • Did you adhere to the code formatting guidelines (TBD)
  • Did you group your changes for easy review, providing meaningful descriptions for each commit?
  • Did you ensure that all files contain the correct copyright header?
  • Did you add documentation for this feature to the release notes directory?

If you did not complete any of these, then please explain below.

@tlwillke
tlwillke merged commit ed5f016 into batched-vector-access Oct 12, 2026
13 checks passed
@tlwillke
tlwillke deleted the repros-744 branch October 12, 2026 21:42
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@MarkWolters MarkWolters Awaiting requested review from MarkWolters
@tlwillke tlwillke Awaiting requested review from tlwillke
@ashkrisk ashkrisk Awaiting requested review from ashkrisk

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@jshook @tlwillke