Lars Volker has posted comments on this change. ( http://gerrit.cloudera.org:8080/11355 )
Change subject: IMPALA-5937: [DOCS] PARQUET_READ_STATISTICS and PARQUET_DICTIONARY_FILTERING ...................................................................... Patch Set 3: (6 comments) http://gerrit.cloudera.org:8080/#/c/11355/3/docs/topics/impala_parquet.xml File docs/topics/impala_parquet.xml: http://gerrit.cloudera.org:8080/#/c/11355/3/docs/topics/impala_parquet.xml@40 PS3, Line 40: helps Not your change, but I'd reword as "allows you to". I don't feel strongly about it though. http://gerrit.cloudera.org:8080/#/c/11355/3/docs/topics/impala_parquet_dictionary_filtering.xml File docs/topics/impala_parquet_dictionary_filtering.xml: http://gerrit.cloudera.org:8080/#/c/11355/3/docs/topics/impala_parquet_dictionary_filtering.xml@44 PS3, Line 44: indicates "controls": Query options will influence Impala's behavior. The profiles will then indicate whether Impala used dictionary filtering, depending on factors like file formats, query predicates, etc. http://gerrit.cloudera.org:8080/#/c/11355/3/docs/topics/impala_parquet_dictionary_filtering.xml@79 PS3, Line 79: Dictionary filtering works for the Parquet dictionaries with less than 40000 values. This is only true if the file was written by Impala < 2.9 and the dictionary has less than 40000 values. Can you clarify? http://gerrit.cloudera.org:8080/#/c/11355/3/docs/topics/impala_parquet_dictionary_filtering.xml@83 PS3, Line 83: We should mention the NumDictFilteredRowGroups field in the profiles that should the number of row groups that were skipped based on dict filtering. http://gerrit.cloudera.org:8080/#/c/11355/3/docs/topics/impala_parquet_read_statistics.xml File docs/topics/impala_parquet_read_statistics.xml: http://gerrit.cloudera.org:8080/#/c/11355/3/docs/topics/impala_parquet_read_statistics.xml@44 PS3, Line 44: indicates controls? http://gerrit.cloudera.org:8080/#/c/11355/3/docs/topics/impala_parquet_read_statistics.xml@53 PS3, Line 53: I think it would be good to point out the caveats here, similar to how we did for dictionary filtering. It should be clear in which cases dict filtering is expected to work, and when it is not. We should also mention the NumStatsFilteredRowGroups field in the profile that shows the number of row groups that were skipped based on stats. -- To view, visit http://gerrit.cloudera.org:8080/11355 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I88fa8c4a64560711251076c50e1695f7f032f9c0 Gerrit-Change-Number: 11355 Gerrit-PatchSet: 3 Gerrit-Owner: Alex Rodoni <[email protected]> Gerrit-Reviewer: Alex Rodoni <[email protected]> Gerrit-Reviewer: Fredy Wijaya <[email protected]> Gerrit-Reviewer: Impala Public Jenkins <[email protected]> Gerrit-Reviewer: Joe McDonnell <[email protected]> Gerrit-Reviewer: Lars Volker <[email protected]> Gerrit-Comment-Date: Tue, 04 Sep 2018 16:50:53 +0000 Gerrit-HasComments: Yes
