njayaram2 commented on a change in pull request #364: WCC: Performance
Improvement
URL: https://github.com/apache/madlib/pull/364#discussion_r273656867
##########
File path: src/ports/postgres/modules/graph/wcc.py_in
##########
@@ -240,27 +250,28 @@ def wcc(schema_madlib, vertex_table, vertex_id,
edge_table, edge_args,
plpy.execute("DROP TABLE IF EXISTS {0}".format(message))
plpy.execute("""
CREATE TEMP TABLE {message} AS
- SELECT {vertex_id}, MIN({component_id}) AS {component_id}
- {select_grouping_cols}
- FROM (
- SELECT {edge_table}.{src} AS {vertex_id},
- {toupdate}.{component_id}
+ SELECT {edge_inverse}.{src} AS {vertex_id},
+ MIN({toupdate}.{component_id}) AS {component_id}
{comma_toupdate_prefixed_grouping_cols}
- FROM {toupdate}, {edge_table}
- WHERE {edge_table}.{dest} = {toupdate}.{vertex_id}
- {edge_to_update_where_condition}
- UNION ALL
+ FROM {toupdate}, {edge_inverse}
+ WHERE {edge_inverse}.{dest} = {toupdate}.{vertex_id}
+ {edge_inverse_to_update_where_condition}
+ GROUP BY {edge_inverse}.{src}
{comma_toupdate_prefixed_grouping_cols}
+ """.format(select_grouping_cols='' if not grouping_cols
+ else ', {0}'.format(grouping_cols),
+ **locals()))
+
+ plpy.execute("""
+ INSERT INTO {message}
SELECT {edge_table}.{dest} AS {vertex_id},
- {toupdate}.{component_id}
+ MIN({toupdate}.{component_id}) AS {component_id}
{comma_toupdate_prefixed_grouping_cols}
FROM {toupdate}, {edge_table}
WHERE {edge_table}.{src} = {toupdate}.{vertex_id}
{edge_to_update_where_condition}
- ) AS t
- GROUP BY {group_by_clause} {vertex_id}
+ GROUP BY {edge_table}.{dest}
{comma_toupdate_prefixed_grouping_cols}
""".format(select_grouping_cols='' if not grouping_cols
- else ', {0}'.format(grouping_cols), group_by_clause=''
- if not grouping_cols else ' {0}, '.format(grouping_cols),
+ else ', {0}'.format(grouping_cols),
**locals()))
plpy.execute("DROP TABLE {0}".format(oldupdate))
Review comment:
As mentioned in the previous comment, can you please drop `edge_inverse`
table here? Please note that it is only created on Greenplum.
----------------------------------------------------------------
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
