bowenli86 commented on a change in pull request #8703:
[FLINK-12807][hive]Support Hive table columnstats related operations in
HiveCatalog
URL: https://github.com/apache/flink/pull/8703#discussion_r293541119
##########
File path:
flink-connectors/flink-connector-hive/src/main/java/org/apache/flink/table/catalog/hive/HiveCatalog.java
##########
@@ -1069,7 +1075,22 @@ public void alterTableStatistics(ObjectPath tablePath,
CatalogTableStatistics ta
@Override
public void alterTableColumnStatistics(ObjectPath tablePath,
CatalogColumnStatistics columnStatistics, boolean ignoreIfNotExists) throws
TableNotExistException, CatalogException {
-
+ try {
+ Table hiveTable = getHiveTable(tablePath);
+ // Set table column stats. This only works for
non-partitioned tables.
+ if (!isTablePartitioned(hiveTable)) {
+
client.updateTableColumnStatistics(HiveCatalogUtil.createTableColumnStats(hiveTable,
columnStatistics.getColumnStatisticsData()));
+ } else {
+ throw new
CatalogException(String.format("Failed to alter partition table column stats of
table %s",
Review comment:
I'm kind of in between throwing a general `CatalogException` and throwing a
`TablePartitionedException`. We've been using specific exception for cases like
this in catalog impl, e.g. adding a partition to a non-partitioned table will
throw a `TableNotPartitionException`. CatalogException is mainly used for
runtime errors like network connection errors, and throwing it here will be a
kind of inconsistent behavior.
What do you guys think?
----------------------------------------------------------------
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
