Github user marmbrus commented on a diff in the pull request:
https://github.com/apache/spark/pull/3752#discussion_r22183385
--- Diff:
sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveMetastoreCatalog.scala ---
@@ -55,8 +56,60 @@ private[hive] class HiveMetastoreCatalog(hive:
HiveContext) extends Catalog with
/** Connection to hive metastore. Usages should lock on `this`. */
protected[hive] val client = Hive.get(hive.hiveconf)
+ // TODO: Use this everywhere instead of tuples or databaseName,
tableName,.
+ /** A fully qualified identifier for a table (i.e., database.tableName)
*/
+ case class TableIdent(database: String, name: String) {
+ def toLowerCase = TableIdent(database.toLowerCase, name.toLowerCase)
+ }
+
+ /** A cache of Spark SQL data source tables that have been accessed. */
+ protected[hive] val cachedDataSourceTables = CacheBuilder.newBuilder()
+ .maximumSize(1000)
+ .build(
+ new CacheLoader[TableIdent, LogicalPlan]() {
+ override def load(in: TableIdent): LogicalPlan = {
+ logDebug(s"Creating new cached data source for $in")
+ val table = client.getTable(in.database, in.name)
+
+ // It does not appear that the ql client for the metastore has a
way to enumerate all the
+ // SerDe properties directly...
+ val method = classOf[Table].getDeclaredMethod("getSerdeInfo")
--- End diff --
My reasoning here was that table properties might have other things that
could conflict with the options that a data source requires. It seems like
SerDe properties are scoped to hold just the options that describe how the
serialization library should read the data (which seems analogous to our data
sources).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
