babelfish-for-postgresql / babelfish_extensions / 18303621061

77%

push

github

Fix precision/scale of numeric columns in SELECT INTO with GROUP BY. (#4121)

1. Issues
If a GROUP BY clause is used in a SELECT statement, the target list during initialisation of TupleDesc will contains

all target columns mentioned in SELECT statement
columns from the GROUP BY clause (In this case column "id")
The columns from GROUP BY clause which is not there in target columns are marked as junk column, which will be filtered out later. During initialisation of TupleDesc we are computing and storing the correct typmod information for all the columns in target list
Later after TupleDesc is initialised, if any of the column is marked as junk, Junk filtration is done which recomputes the entire TupleDesc which leads to losing the typmod information of avgValue column.

2. Changes made to fix the issues
Earlier we used to return typmod as -1 if plan was NULL for T_var node whereas correct typmod was preserved in var->vartypmod. So we fixed that portion in thi PR.

Issues Resolved
Task: BABEL-6081

Signed-off-by: yashneet vinayak 
yashneet@amazon.com

6 of 6 new or added lines in 1 file covered. (100.0%)

51689 of 67445 relevant lines covered (76.64%)

383346.65 hits per line