feat: shared ScanScheduler, FileScheduler cache, and V1 reader cache on Dataset#3

Open

jeremywgleeson wants to merge 14 commits intomainfrom

devin/1774288046-cache-fragment-readers

Commits on Mar 23, 2026

feat: store shared ScanScheduler on Dataset for reuse across fragment reads

devin-ai-integration[bot]
and
jeremywgleeson
committed
style: fix rustfmt formatting for shared scan scheduler tuple

devin-ai-integration[bot]
and
jeremywgleeson
committed
feat: add FileScheduler cache on Dataset to avoid re-opening S3 files per request

devin-ai-integration[bot]
and
jeremywgleeson
committed
style: fix import ordering for rustfmt compliance

devin-ai-integration[bot]
and
jeremywgleeson
committed
style: fix rustfmt formatting for FileScheduler cache code

devin-ai-integration[bot]
and
jeremywgleeson
committed
perf: add S3 call counting and scheduler timing instrumentation

devin-ai-integration[bot]
and
jeremywgleeson
committed
perf: add S3 byte counting and V1/V2 format logging for profiling

devin-ai-integration[bot]
and
jeremywgleeson
committed
style: fix rustfmt formatting for V1/V2 format logging

devin-ai-integration[bot]
and
jeremywgleeson
committed
perf: add V1 (CloudObjectReader) S3 call counting for profiling

devin-ai-integration[bot]
and
jeremywgleeson
committed
style: fix import ordering in lib.rs

devin-ai-integration[bot]
and
jeremywgleeson
committed
feat: re-export S3 profiling counters through lance crate

devin-ai-integration[bot]
and
jeremywgleeson
committed
perf: cache V1 PreviousFileReader instances on Dataset to avoid per-request S3 reader creation

devin-ai-integration[bot]
and
jeremywgleeson
committed

Commits on Mar 24, 2026