rdblue commented on a change in pull request #3954:
URL: https://github.com/apache/iceberg/pull/3954#discussion_r790354893
##########
File path: core/src/main/java/org/apache/iceberg/BaseUpdatePartitionSpec.java
##########
@@ -124,6 +124,17 @@ public BaseUpdatePartitionSpec addField(Term term) {
return addField(null, term);
}
+ private BaseUpdatePartitionSpec rewriteDeleteAndAddField(
+ PartitionField existing, String name, Pair<Integer, Transform<?, ?>>
sourceTransform) {
+ deletes.remove(existing.fieldId());
+ if (existing.name().equals(name)) {
+ return this;
+ } else {
+ String newName = name == null ? sourceTransform.second().toString() :
name;
+ return renameField(existing.name(), newName);
Review comment:
This uses the transform's string representation as the partition field
name, but that's not how we generate partition field names. In fact, the result
is that we have a partition field name for bucket transforms that is invalid or
must be quoted in most systems (`"bucket[16]"`).
Rather than generating the name, I think that this should consider a `null`
name a no-op. That is if I remove `shard` and add `bucket(16, id)` then the
result should still be named shard -- there's no reason to drop the name and we
avoid breaking metadata tables needlessly.
However, if the name was explicitly set, then drop and and should act as a
rename.
In short, I think it should be this:
```java
if (name == null || existing.name().equals(name)) {
return this;
} else {
return renameField(existing.name(), name);
}
```
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
