[GitHub] [druid] abhishekagarwal87 commented on a change in pull request #10366: Proposed changes for making joins cacheable

[GitBox <gi...@apache.org>]

abhishekagarwal87 commented on a change in pull request #10366:
URL: https://github.com/apache/druid/pull/10366#discussion_r487695326



##########
File path: processing/src/main/java/org/apache/druid/segment/join/Joinables.java
##########
@@ -118,6 +126,47 @@ public static boolean isPrefixedBy(final String columnName, final String prefix)
     );
   }
 
+  /**
+   * Compute a cache key prefix for data sources that is to be used in segment level and result level caches. The
+   * data source can either be base (clauses is empty) or RHS of a join (clauses is non-empty). In both of the cases,
+   * a non-null cache is returned. However, the cache key is null if there is a join and some of the right data sources
+   * participating in the join do not support caching yet
+   *
+   * @param dataSourceAnalysis
+   * @param joinableFactory
+   * @return
+   */
+  public static Optional<byte[]> computeDataSourceCacheKey(
+      final DataSourceAnalysis dataSourceAnalysis,
+      final JoinableFactory joinableFactory
+  )
+  {
+    final CacheKeyBuilder keyBuilder;
+    final List<PreJoinableClause> clauses = dataSourceAnalysis.getPreJoinableClauses();
+    if (clauses.isEmpty()) {
+      keyBuilder = new CacheKeyBuilder(REGULAR_OPERATION);
+    } else {
+      keyBuilder = new CacheKeyBuilder(JOIN_OPERATION);
+      for (PreJoinableClause clause : clauses) {
+        if (!clause.getCondition().canHashJoin()) {

Review comment:
       I was thinking to keep it here so that I could just pass the `data-source` inside the `computeJoinCacheKey`.  The engine for hash join works outside the boundaries of joinable factories. so it felt simpler to just keep it here and joinable factories just worry about the data source. The other cache info from the condition is derived here in this class as well (original expression etc) 




----------------------------------------------------------------
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:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For additional commands, e-mail: commits-help@druid.apache.org