Github user njayaram2 commented on a diff in the pull request:
https://github.com/apache/madlib/pull/315#discussion_r213792484
--- Diff: src/ports/postgres/modules/knn/knn.py_in ---
@@ -243,6 +237,23 @@ def knn(schema_madlib, point_source,
point_column_name, point_id,
label_out = ""
comma_label_out_alias = ""
+ # CTAS temporary table from the main table(s) to select the
+ # given expression from.
+
+ point_source_temp_table = unique_string(desp='point_source')
+ test_source_temp_table = unique_string(desp='test_source')
+ point_col_name_temp = unique_string(desp='point_col_name_temp')
+ test_col_name_temp = unique_string(desp='test_col_name_temp')
+
+ plpy.execute("""
+ create table {point_source_temp_table} as
+ (select {point_id} , {point_column_name} as
{point_col_name_temp} , {label_column_name} from {point_source})"""
+ .format(**locals()))
+ plpy.execute("""
+ create table {test_source_temp_table} as
+ (select {test_id}, {test_column_name} as
{test_col_name_temp} from {test_source})"""
+ .format(**locals()))
--- End diff --
Did you mean to create a temp table for these?
Either way, we don't have to create a table for these. We could directly
run them as subqueries instead.
---
