framework/resources/Functional/joins/README

DATA:
----

j1 - parquet table generated using CTAS from j1.csv file in drill
	-- 10,000 rows
	-- 99 parquet files 
	-- 5000 unique values
	-- 200 NULL values
	-- all columns are of optional data type

j2 - parquet table generated using CTAS from j2.csv file in drill
	-- 10 rows
	-- 1 parquet file 
	-- All values are unique and have corresponding values in j1
	-- 1 NULL value
	-- all columns are of optional data type

j3 - parquet table generated using CTAS from j3.csv file in drill
	-- 100 rows
	-- 1 parquet file 
	-- All values are unique and have corresponding values in j1
	-- NO NULL values
	-- all columns are of required data type

j4 - parquet table generated using CTAS from j4.csv file in drill
        -- 1000 rows
        -- 1 parquet file
        -- All values are unique and have corresponding values in j1
        -- NO NULL values
        -- all columns are of required data type

cast_tbl_1 - table used for implicit/explicit cast testing
cast_tbl_2 - table used for implicit/explicit cast testing

test - table used for coalesce testing, sparse data with lots of nulls

TODO:
	* convert cast_tbl_1 to required type
	* Do we want to run ALL the tests against views ?
              Created views with explicit types on top of j1[234] files and ran all the tests with 0.8.0 rc2
              Only one bug was found: drill-2627
	* Add tests for running with hash/merge join (alter session ...) if "permutations" feature is not added to
          framework by 04/10/15 : see individual README files for each suite
	* Add plan testing for order_by queries, after drill-2655 is fixed
	* Add some joins with correlation and inequality, was not in the original plan
	* Check in sql scripts that we used to generate postgres tables
	* Check in sql scripts that we used to generate drill tables
	* When decimal type is reintroduced, it will be sufficient to regenerate all the data again 
          and rerun all the tests. Columns: d9,d18,d28,d38 were planned to be of decimal type, currently they are all double.