rdblue commented on a change in pull request #351: Extend Iceberg with a way to
overwrite files for eager updates/deletes
URL: https://github.com/apache/incubator-iceberg/pull/351#discussion_r313647841
##########
File path: core/src/main/java/org/apache/iceberg/OverwriteData.java
##########
@@ -88,6 +121,40 @@ public OverwriteFiles validateAddedFiles() {
}
}
+ if (conflictingAppendsRowFilter != null) {
+ PartitionSpec spec = writeSpec();
+ Expression inclusiveExpr =
Projections.inclusive(spec).project(conflictingAppendsRowFilter);
+ Evaluator inclusive = new Evaluator(spec.partitionType(), inclusiveExpr);
+
+ List<DataFile> newFiles = collectNewFiles(base);
+ for (DataFile newFile : newFiles) {
+ // we do partition-level conflict resolution right now
+ // we can enhance it by leveraging column stats and MetricsEvaluator
+ ValidationException.check(
+ !inclusive.eval(newFile.partition()),
+ "A conflicting file was appended that matches filter '%s': %s",
+ conflictingAppendsRowFilter, newFile.path());
+ }
+ }
+
return super.apply(base);
}
+
+ private List<DataFile> collectNewFiles(TableMetadata meta) {
+ List<DataFile> newFiles = new ArrayList<>();
+
+ Long currentSnapshotId = meta.currentSnapshot() == null ? null :
meta.currentSnapshot().snapshotId();
+ while (currentSnapshotId != null &&
!currentSnapshotId.equals(readSnapshotId)) {
+ Snapshot currentSnapshot = meta.snapshot(currentSnapshotId);
+
+ if (currentSnapshot == null) {
+ throw new ValidationException("Cannot find snapshot %d. Was it
expired?", currentSnapshotId);
Review comment:
If `meta.currentSnapshot` is null, this will print "Cannot find snapshot
null". I think it would be better to start the method with a precondition that
checks that meta has a current snapshot and fails with a better error message
if it doesn't.
Also, there are a few reasons why the snapshot can't be found. The base
snapshot may have been expired, but it also could have been rolled back. I
think this should just state "Cannot determine history between base snapshot %s
and current %s".
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
