Conversation
Rebuilds the V4 Adaptive Metadata Tree prototype on current main using the native scan planner for reads and a minimal flat-tree write path. Reads: - Add ScanTaskPlanner, which plans FileScanTasks directly from a V4 root manifest of TrackedFile entries. - Wire DataTableScan and BaseDistributedDataScan to plan via ScanTaskPlanner for format version 4. Writes: - SnapshotProducer.applyV4 writes a Parquet root manifest for v4 tables: the appended data files are inlined as DATA TrackedFile entries and data entries from the parent snapshot's root manifest are carried forward. Replaces the Avro manifest list for v4. - BaseSnapshot reads a Parquet root manifest. - Make the TrackedFileAdapters content-file adapters Serializable so planner output can be shipped to Spark executors. Validated end to end: core round trip (TestV4FlatTreeReadWrite), Spark TestV4ReadEndToEnd, and an interactive spark-sql session.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Rebuilds the V4 Adaptive Metadata Tree prototype on current main using the native scan planner for reads and a minimal flat-tree write path.
Reads:
Writes:
SnapshotProducer.applyV4 writes a Parquet root manifest for v4 tables: the appended data files are inlined as DATA TrackedFile entries and data entries from the parent snapshot's root manifest are carried forward. Replaces the Avro manifest list for v4.
BaseSnapshot reads a Parquet root manifest.
Make the TrackedFileAdapters content-file adapters Serializable so planner output can be shipped to Spark executors.
Validated end to end: core round trip (TestV4FlatTreeReadWrite), Spark TestV4ReadEndToEnd, and an interactive spark-sql session.