----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/27640/#review60031 -----------------------------------------------------------
Hi, I know this is a combination of patches from Suhas and Chao, but I left all some comments/questions here for the overall patch. ql/src/java/org/apache/hadoop/hive/ql/optimizer/physical/SparkMapJoinResolver.java <https://reviews.apache.org/r/27640/#comment101369> Please use proper javadoc notation for your javadocs. ql/src/java/org/apache/hadoop/hive/ql/optimizer/physical/SparkMapJoinResolver.java <https://reviews.apache.org/r/27640/#comment101365> Can you elaborate why we need this assumption? This may not be true in all cases. ql/src/java/org/apache/hadoop/hive/ql/optimizer/physical/SparkMapJoinResolver.java <https://reviews.apache.org/r/27640/#comment101373> We need to handle the case of where parent is the mapwork of big-table. Basically if we are walking from big-table to a reduce work of mapjoin, we should not replace RS with HTS. (it can happen for example, if big-table is the result from a group by). Only work of small-tables should get replaced. ql/src/java/org/apache/hadoop/hive/ql/optimizer/physical/SparkMapJoinResolver.java <https://reviews.apache.org/r/27640/#comment101371> Let's just get mapJoinOp and check if its null in one call. (Can get rid of one of the methods). ql/src/java/org/apache/hadoop/hive/ql/optimizer/physical/SparkMapJoinResolver.java <https://reviews.apache.org/r/27640/#comment101363> Fix indent here. ql/src/java/org/apache/hadoop/hive/ql/optimizer/physical/SparkMapJoinResolver.java <https://reviews.apache.org/r/27640/#comment101366> Please fix all these tabs, and make sure to indent them properly. ql/src/java/org/apache/hadoop/hive/ql/optimizer/physical/SparkMapJoinResolver.java <https://reviews.apache.org/r/27640/#comment101368> Are you sure we dont need to initialize the HTSOperator's values like it does in LocalMapJoinProcFactory? ql/src/java/org/apache/hadoop/hive/ql/optimizer/physical/SparkMapJoinResolver.java <https://reviews.apache.org/r/27640/#comment101367> Nit: 'parentOps' is probably a better name. - Szehon Ho On Nov. 5, 2014, 8:29 p.m., Suhas Satish wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/27640/ > ----------------------------------------------------------- > > (Updated Nov. 5, 2014, 8:29 p.m.) > > > Review request for hive, Chao Sun, Jimmy Xiang, Szehon Ho, and Xuefu Zhang. > > > Repository: hive-git > > > Description > ------- > > This replaces ReduceSinks with HashTableSinks in smaller tables for a > map-join. But the condition check field to detect map-join is actually being > set in CommonJoinResolver, which doesnt exist yet. We need to decide where is > the right place to populate this field. > > > Diffs > ----- > > > ql/src/java/org/apache/hadoop/hive/ql/optimizer/physical/SparkMapJoinResolver.java > PRE-CREATION > ql/src/java/org/apache/hadoop/hive/ql/parse/spark/SparkCompiler.java > 795a5d7 > > Diff: https://reviews.apache.org/r/27640/diff/ > > > Testing > ------- > > > Thanks, > > Suhas Satish > >
