Title: fix: scan misaligned per-column chunk grids (#221) by dfa1 · Pull Request #223 · dfa1/vortex-java · GitHub
Open Graph Title: fix: scan misaligned per-column chunk grids (#221) by dfa1 · Pull Request #223 · dfa1/vortex-java
X Title: fix: scan misaligned per-column chunk grids (#221) by dfa1 · Pull Request #223 · dfa1/vortex-java
Description: Fixes #221, found by Raincloud triage round 3 (#205): ScanIterator.buildChunks rejected files whose columns use different chunk grids ("mixed per-column chunking beyond 1-vs-N is not supported") — files the Rust reader handles fine. Approach mirrors the Rust reference (vortex-layout/src/scan/split_by.rs + register_splits down the layout tree): scan windows are the sorted-deduped union of every column's chunk boundaries. For each window, each column advances a monotonic cursor to its covering chunk and slices to the window. This one model subsumes both failure tiers — nested boundaries (emotions) and truly disjoint grids (beijing) — with no tier-specific code. No regression to the fast paths: isWholeFlat gates a window that equals a whole chunk straight into that chunk's arena (aligned N-vs-N and 1-vs-N keep per-chunk zero-copy, no slice wrapper). A coarse chunk spanning several windows is decoded once into a shared arena and sliced zero-copy per window. Validation: New ScanIteratorChunkGridTest (aligned, 1-vs-N, nested [8,8,8,4]-vs-[2×14], disjoint [3,3,2]-vs-[4,4], empty). Reader 1143 tests green; full verify -pl integration -am 287 ITs green (no scan regressions). Real files: uci-beijing-multi-site-air-quality (7.5M cells) and emotions-dataset-for-nlp (833k cells) both export and match the Parquet oracle exactly. Matrix: both gap:221 → ok (30 ok / 0 gaps); docs + CHANGELOG updated. Reviewer note (from the agent): coarse covering flats are cached for the iterator's lifetime in a shared arena — correctness and decode-once are satisfied; per-chunk arena eviction on a very wide disjoint file is a possible future memory optimization, not a correctness issue. Closes #221 🤖 Generated with Claude Code
Open Graph Description: Fixes #221, found by Raincloud triage round 3 (#205): ScanIterator.buildChunks rejected files whose columns use different chunk grids ("mixed per-column chunking beyond 1-vs-N is not supported...
X Description: Fixes #221, found by Raincloud triage round 3 (#205): ScanIterator.buildChunks rejected files whose columns use different chunk grids ("mixed per-column chunking beyond 1-vs-N is not suppo...
Opengraph URL: https://github.com/dfa1/vortex-java/pull/223
X: @github
Domain: github.com
| route-pattern | /:user_id/:repository/pull/:id/commits/:range(.:format) |
| route-controller | pull_requests |
| route-action | commits |
| fetch-nonce | v2:5bd1f5b3-6903-dd64-8a20-8fa82694ca56 |
| current-catalog-service-hash | ae870bc5e265a340912cde392f23dad3671a0a881730ffdadd82f2f57d81641b |
| request-id | C6DE:54847:4FCFD4B:6A69E37:6A655898 |
| html-safe-nonce | 50928c05b1defb754711fdb29db659adf718fe70ab5d7f710bedfebd658447ea |
| visitor-payload | eyJyZWZlcnJlciI6IiIsInJlcXVlc3RfaWQiOiJDNkRFOjU0ODQ3OjRGQ0ZENEI6NkE2OUUzNzo2QTY1NTg5OCIsInZpc2l0b3JfaWQiOiI0NTk0MTM0MDc2ODk2OTMzMzYiLCJyZWdpb25fZWRnZSI6ImlhZCIsInJlZ2lvbl9yZW5kZXIiOiJpYWQifQ== |
| visitor-hmac | d6574608b94a0342858da848527e857819e3384f12a1d0b5d487153f279dc410 |
| hovercard-subject-tag | pull_request:4004181792 |
| github-keyboard-shortcuts | repository,pull-request-list,pull-request-conversation,pull-request-files-changed,copilot |
| google-site-verification | Apib7-x98H0j5cPqHWwSMm6dNU4GmODRoqxLiDzdx9I |
| octolytics-url | https://collector.github.com/github/collect |
| analytics-location | / |
| fb:app_id | 1401488693436528 |
| apple-itunes-app | app-id=1477376905, app-argument=https://github.com/dfa1/vortex-java/pull/223/commits/023041f92dd75eaf2e0314ba8a4de9107f238a29 |
| twitter:image | https://avatars.githubusercontent.com/u/5798698?s=400&v=4 |
| twitter:card | summary_large_image |
| og:image | https://avatars.githubusercontent.com/u/5798698?s=400&v=4 |
| og:image:alt | Fixes #221, found by Raincloud triage round 3 (#205): ScanIterator.buildChunks rejected files whose columns use different chunk grids ("mixed per-column chunking beyond 1-vs-N is not supported... |
| og:site_name | GitHub |
| og:type | object |
| hostname | github.com |
| expected-hostname | github.com |
| None | 52c76df668885aaff23b50bdca1fa1ea44ac9c1553e888ebc70ff1e4daa4625b |
| turbo-cache-control | no-preview |
| diff-view | unified |
| go-import | github.com/dfa1/vortex-java git https://github.com/dfa1/vortex-java.git |
| octolytics-dimension-user_id | 5798698 |
| octolytics-dimension-user_login | dfa1 |
| octolytics-dimension-repository_id | 1249412947 |
| octolytics-dimension-repository_nwo | dfa1/vortex-java |
| octolytics-dimension-repository_public | true |
| octolytics-dimension-repository_is_fork | false |
| octolytics-dimension-repository_network_root_id | 1249412947 |
| octolytics-dimension-repository_network_root_nwo | dfa1/vortex-java |
| turbo-body-classes | logged-out env-production page-responsive full-width |
| disable-turbo | false |
| browser-stats-url | https://api.github.com/_private/browser/stats |
| browser-errors-url | https://api.github.com/_private/browser/errors |
| release | 309153364422b3c499922d1a2a6404910a58ed8e |
| ui-target | full |
| theme-color | #1e2327 |
| color-scheme | light dark |
Links:
Viewport: width=device-width