feat: Incremental append scan#2337
Open
xanderbailey wants to merge 32 commits into
Open
Conversation
xanderbailey
force-pushed
the
xb/incremental_read
branch
from
e88d444 to
11cce80
Compare
xanderbailey
commented
Apr 16, 2026
Comment on lines
+36
to
+43
| /// Filter applied to each [`ManifestFile`] before fetching it. | ||
| /// Returns `true` to include the manifest, `false` to skip it. | ||
| pub(crate) type ManifestFileFilter = Arc<dyn Fn(&ManifestFile) -> bool + Send + Sync>; | ||
|
|
||
| /// Filter applied to each manifest entry after loading a manifest. | ||
| /// Returns `true` to include the entry, `false` to skip it. | ||
| pub(crate) type ManifestEntryFilter = Arc<dyn Fn(&ManifestEntryRef) -> bool + Send + Sync>; | ||
|
|
Contributor
Author
There was a problem hiding this comment.
I think this is a nice way to inject these filters and should extend to other future scans
xanderbailey
commented
Apr 16, 2026
| pub manifest_entry_filter: Option<ManifestEntryFilter>, | ||
| } | ||
|
|
||
| impl std::fmt::Debug for PlanContext { |
Contributor
Author
There was a problem hiding this comment.
ManifestFileFilter and ManifestEntryFilter can't be debug
CTTY
reviewed
Apr 16, 2026
| pub(crate) struct AppendSnapshotSet { | ||
| /// Snapshot IDs in the range | ||
| snapshot_ids: HashSet<i64>, | ||
| } |
Collaborator
There was a problem hiding this comment.
Contributor
Author
There was a problem hiding this comment.
ancestors_between is used here. We need to add validation and operation-type checking on top of it, we're not re-implementing the traversal. Does that make sense?
xanderbailey
commented
Apr 26, 2026
| /// | ||
| /// Use [`Table::incremental_append_scan`] or | ||
| /// [`Table::incremental_append_scan_inclusive`] to create an instance. | ||
| pub struct IncrementalAppendScanBuilder<'a> { |
Contributor
Author
There was a problem hiding this comment.
There is duplication here between the scan builders which I've left for now until maybe we have a 3rd variant and we can pick out the common patterns a bit better.
Contributor
Author
|
@CTTY would you be open to giving a review here? |
xanderbailey
force-pushed
the
xb/incremental_read
branch
from
3a4f875 to
489084d
Compare
Resolve conflicts in scan/context.rs and scan/mod.rs by combining the incremental-scan work (entry_filter, build_table_scan/ScanConfig refactor) with main's name_mapping support and metadata-column validation. Adapt incremental-scan test fixtures to the updated ManifestWriterBuilder::new and ManifestListWriter::v2 signatures from the encryption changes.
…er tests Arc<dyn ExecutionPlan> supports downcast_ref directly; the extra as_any() call resolved to the unrelated as-any crate trait (not in scope), which broke compilation and cascaded into spurious deref errors on the matched i64 fields. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Match the Java and PyIceberg implementations: an incremental append scan now projects onto the table's current schema rather than the to-snapshot's schema. Rows written under an older schema within the range are read against the current schema, so newer columns become NULL. Adds a schema_override to ScanConfig (None for regular scans, which keeps the correct per-snapshot schema for time-travel; current schema for incremental scans) and a stale-schema deep-history fixture plus a test asserting the projected schema is the current one. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Incorporate the substantive reviewer feedback from the py-iceberg incremental append scan PR (apache/iceberg-python#3512): - Document that this is not a CDC/net-changes/changelog scan and that compaction (replace) output is skipped, so appended rows are never double-counted (kevinjqliu). - Document the ancestors_between footgun: a from-snapshot that is not in the lineage silently yields all ancestors of the to-snapshot, so callers must validate the lineage (kevinjqliu). - Add an explicit compaction (replace) end-to-end test proving the rewritten file is not picked up (rambleraptor). Deep-history fixture variants now share one helper and mutate the parsed serde_json::Value structurally by field, instead of fragile raw string substitution on the shared testdata file. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the serde_json::Value mutation with dedicated templated metadata files for the stale-schema and compaction deep-history fixtures, matching the repo's existing testdata pattern. The three fixtures now share a filename-parameterized helper that renders the manifest-list paths. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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
2 participants
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.
Which issue does this PR close?
What changes are included in this PR?
Adds incremental append scan support to iceberg-rust, allowing users to read only newly added data files between two snapshots. This is the Rust equivalent of Java's
BaseIncrementalAppendScan.Are these changes tested?
Yes
Java comparison notes:
AppendSnapshotSet::buildmirrors Java'sBaseIncrementalAppendScan.snapshotsBetween()— only APPEND snapshots are collected, non-APPEND are silently skippedfrom_snapshotsemantics match Java's defaultIncrementalAppendScanbehavioradded_snapshot_id) and entry filtering (status == ADDED,snapshot_idin range) matchBaseIncrementalAppendScan.doPlanFiles()in Java