Union of crazy on 3 tables

Similar Questions

• With the help of SCORE on top of view with UNION

  Hi guys,.

  I explain what I'm trying to do it quickly:

  2 tables: table1 and table2 with the same structure and have both a multi_column_datastore ctxsys.context.

  1 view: View1

  Select * from table1

  Union

  Select * from table2

  If I run:

  Select * from View1 WHERE contains (view1. Column1, '% textext %', 1) > 0;

  It works fine, I get the correct result.

  If I try to use the PARTITION function, I got an error:

  Select * from View1 WHERE contains (view1. Column1, '% textext %', 1) > 0 ORDER by SCORE (1);

  ORA-29921: auxiliary operator not supported with query to configure block

  I understand the problem is in the UNION inside the view, is it possible to keep work, filtering the VIEW?

  Thanks in advance

  There is no score in the view, so you cannot reference the score when you query the view.  In order to put the note in the view, you need a contains the clause, which requires a value.  A method to do this is to use sys_context.  Please see the reproduction of the problem and solution below.

  Scott@orcl12c >-reproduction of the problem:

  Scott@orcl12c > create the table1 table:

  2 (column1 varchar2 (30))

  3.

  Table created.

  Scott@orcl12c > insert into table1 values ('textext")

  2.

  1 line of creation.

  Scott@orcl12c > create table table2

  2 (column1 varchar2 (30))

  3.

  Table created.

  Scott@orcl12c > insert into table2 values ('textext")

  2.

  1 line of creation.

  Scott@orcl12c > start

  2 ctx_ddl.create_preference ('test_ds', 'multi_column_datastore');

  3 ctx_ddl.set_attribute ('test_ds', 'columns', "column1");

  4 end;

  5.

  PL/SQL procedure successfully completed.

  Scott@orcl12c > create index table1_idx on table1 (column1)

  2 indextype is ctxsys.context

  3 parameters ("test_ds of the data store")

  4.

  The index is created.

  Scott@orcl12c > create index table2_idx on the table2 (column1)

  2 indextype is ctxsys.context

  3 parameters ("test_ds of the data store")

  4.

  The index is created.

  Scott@orcl12c > create or replace view View1

  2 as

  3 select * from table1

  4 union

  5 select * from table2

  6.

  Created view.

  Scott@orcl12c > select * from View1 where contains (view1.column1, '% textext %', 1) > 0

  2.

  COLUMN1

  ------------------------------

  textext

  1 selected line.

  Scott@orcl12c > select * from View1 where contains > 0 (view1.column1,'%textext%',1) order by score (1)

  2.

  Select * from View1 where contains > 0 (view1.column1,'%textext%',1) order by score (1)

  *

  ERROR on line 1:

  ORA-29921: auxiliary operator not supported with query to configure block

  Scott@orcl12c >-solution:

  Scott@orcl12c > create or replace view View1

  2 as

  3. Select the partition (1) score, table1.* from table1

  4 where contains (table1. Column1, sys_context ('text_query', 'query_value'), 1) > 0

  5 union

  6 select score partition (1), table2.* from table2

  7 where contains (table2.column1, sys_context ('text_query', 'query_value'), 1) > 0

  8.

  Created view.

  Scott@orcl12c > create or replace context text_query using text_proc

  2.

  Context that is created.

  Scott@orcl12c > create or replace procedure text_proc

  2 (p_val in varchar2)

  3 as

  4 start

  5 dbms_session.set_context ('text_query', "query_value", p_val);

  6 end text_proc;

  7.

  Created procedure.

  Scott@orcl12c > text_proc exec ('% textext %')

  PL/SQL procedure successfully completed.

  Scott@orcl12c > set autotrace on explain

  Scott@orcl12c > select * from View1 by score

  2.

  MARK THE COLUMN1

  ---------- ------------------------------

  3 textext

  1 selected line.

  Execution plan

  ----------------------------------------------------------

  Hash value of plan: 4090246122

  ----------------------------------------------------------------------------------------------

  | ID | Operation | Name | Lines | Bytes | Cost (% CPU). Time |

  ----------------------------------------------------------------------------------------------

  |   0 | SELECT STATEMENT |            |     2.    60.     8 (0) | 00:00:01 |

  |   1.  SORT ORDER BY |            |     2.    60.     8 (0) | 00:00:01 |

  |   2.   VIEW                          | VIEW1.     2.    60.     8 (0) | 00:00:01 |

  |   3.    UNIQUE FATE |            |     2.    58.     8 (50) | 00:00:01 |

  |   4.     UNION-ALL |            |       |       |            |          |

  |   5.      TABLE ACCESS BY INDEX ROWID | TABLE1.     1.    29.     4 (0) | 00:00:01 |

  |*  6 |       DOMAIN INDEX | TABLE1_IDX |       |       |     4 (0) | 00:00:01 |

  |   7.      TABLE ACCESS BY INDEX ROWID | TABLE2.     1.    29.     4 (0) | 00:00:01 |

  |*  8 |       DOMAIN INDEX | TABLE2_IDX |       |       |     4 (0) | 00:00:01 |

  ----------------------------------------------------------------------------------------------

  Information of predicates (identified by the operation identity card):

  ---------------------------------------------------

  6 - access("CTXSYS".") CONTAINS "(" TABLE1".»)" Column1', SYS_CONTEXT ('text_query ',' query_v)

  Alou '), 1) > 0)

  8 - access("CTXSYS".") CONTAINS "(" TABLE2".»)" Column1', SYS_CONTEXT ('text_query ',' query_v)

  Alou '), 1) > 0)

  Note

  -----

  -the dynamic statistics used: dynamic sampling (level = 2)

  Scott@orcl12c >

• SQL INTO union-clause

  SQL help

  If any of the queries Union returns a value, I would like to mark as closed status. How can I put the INTO clause in this case because of two queries. Thank you

  BEGIN
  
  SELECT 1

  FROM TableA your
  WHERE tA.col1 = 13
  AND tA.col2 = '100'
  UNION
  SELECT 1
  TB from TableA
  WHERE tb.col2 = 'y '.
  AND tb.col3 = - 123
  AND tb.col4 IS NOT NULL;

  status_cd: = 'closed ';

  EXCEPTION
  WHEN NO_DATA_FOUND THEN
  status_cd: = 'Unknown ';
  
  END;

  Hello

  Here's one way:

  BEGIN

  SELECT 'closed '.

  IN status_cd

  FROM tablea

  WHERE ((col1 = 13

  AND col2 = '100'

  )

  OR (col2 = 'y')

  AND col3 = - 123

  AND col4 IS NOT NULL

  )

  )

  AND ROWNUM = 1;

  EXCEPTION

  WHEN NO_DATA_FOUND

  THEN

  status_cd: = 'Unknown ';

  END;

• create table physical use of select in the repository

  Hi gurus,
  
  We can create physical table in OBIEE 11.1.1.6 repository with stored procedure and select?
  How is the right syntax?
  
  Thank you very much
  
  JOE

  Hello

  Yes. just select and put it under layers of physics
  for example,.
  Select field1, field2. field_n
  tables
  UNION
  Select field1, field2. field_n
  tables;

  http://gerardnico.com/wiki/dat/OBIEE/opaque_view
  http://www.clearpeaks.com/blog/Oracle-BI-EE-11gusing-select_physical-in-OBIEE-11g
  http://allaboutobiee.blogspot.com/2012/05/OBIEE-11g-deployundeploy-view-in.html

  Thank you
  Deva

• Several flat_files through the external table with only the common columns of loading

  Hi, I have 50 flat files and each of them have some columns (fields) common and I need to load only the fields that are common to an external Table. Is any chance to do it with education unique external table. Or I need to load all flat_files at separate tables and then with the ETG and UNION load them only one table.

  If the page size for all the files are different, I think that your only option would be to define different external tables and create a view that joins all the.

  HTH
  Srini

• Alternative and improvement of the union query?

  Hello
  
  I have 2 sets of similar data together with a union query.
  
  Simplified example

  *Table A*
  Name Dept
  Ben    1
  Amy   3
  Ken    3
  
  *Table B*
  Name Dept
  Jim    1
  John  2
  Sue   3
  
  select Name, Dept
  from table A
  UNION
  select Name, Dept
  from table B

  I want to do is to ONLY include data from the second part of the union query, where it is associated with the first part of the union query data.
  
  For example

  select Name, Dept
  from table A
  UNION
  select Name, Dept
  from table B
  where Dept in (select distinct Dept from A)

  That's why John, who works in the Dept 2 would not be included in the results.
  
  Is there a better or more effective way to achieve that my example using the subquery in the where clause?
  
  Thank you very much
  Hazel

  Hi, Hazel,

  What you posted is as good as you can get without changing tables.

  Do not say SEPARATE in the IN-under-query:

  ...
  where   Dept in (  select  Dept
                     from    A
                  )
  

  will get the same results, maybe a little faster.

• Summary of the Union query

  Hi all
  I'm trying yo build a union query, in the simple table view for the sum, I get the following error
  
  Error codes: OPR4ONWY:U9IM8TAC:OI2DL65P
  State: HY000. Code: 10058. [NQODBC] [SQL_STATE: HY000] [nQSError: 10058] A general error occurred. [nQSError: 42038] A REPORT of a total amount of function must be in a worksheet query block. (HY000)
  
  can someone give the solution for this
  
  Thanks in advance.

  Hello

  A check the option based on the Total (if any) report for the totals or subtotals, you applied.

  This option is available after you click the sigma button

• Join the nearest date "not used".

  I need to join the following tables by the next date "not previously signed" the nearest:

  start_table:

  st_id	start_dt
  start0	31-dec-2014
  Start1	January 3, 2015
  Start2	January 5, 2015
  start3	January 7, 2015
  Debut4	8 January 2015
  Debut5	January 14, 2015

  end_table:

  end_id	end_dt
  end0	January 1, 2015
  End1	January 2, 2015
  End2	January 13, 2015
  End3	January 15, 2015
  bout4	January 17, 2015
  end5	19 January 2015
  end6	January 20, 2015

  Result:

  st_id	end_id
  start0	end0
  Start1	End2
  Start2	End3
  start3	bout4
  Debut4	end5
  Debut5	end6

  start0 joined end0, because the closer to the date of the next 31-dec-2014 is 1 January 2015

  Start1 joined end2, because the closer to the date of the next January 3, 2015 is January 13, 2015.

  Start2 joined end3, because as well as the date of the next more close January 5, 2015 is 13 January 2015, this is already accompanied Start1, so she joined next January 15, 2015.

  start3 joined bout4, because the date of the next although most close January 7, 2015 is 13 January 2015, this one is already joined by Start1, earliest date is January 15, 2015, but is also already joined by start2, then he joined the next available date January 17, 2015.

  Database: 11g

  Thanks in advance

  [UPDATE: changed line 37 to add "or cnt > = 0".] [This covers cases where there are more departures that ends at first.]

  Jiri.Machotka - Oracle wrote:

  I found a non recursive algorithm for this problem...

  I came up with something similar: no recursion, no joins, each table read only once.

  1. UNION ALL marked times 1 tables, with lines of departure and end marked lines - 1
  2. Order by date (first lines) and get a combination of 1's and - 1's.
     so when there are too many lines to end, the sum is negative.
  3. Get the previous cumulative minimum! Then take that end with a sum running lines, at less than the previous minimum.
  4. The remaining lines will have an end of line for each line of departure. Now number lines starting from 1 to N and the lines at the end of 1 to N, then match lines start and end in pairs. I use PIVOT to do this.

  WITH end_table (end_id, end_dt) AS (
    select 'end0',to_date('01-jan-2015','dd-mon-yyyy') from dual union all
    select 'end1',to_date('02-jan-2015','dd-mon-yyyy') from dual union all
    select 'end2',to_date('13-jan-2015','dd-mon-yyyy') from dual union all
    SELECT 'end3',to_date('15-jan-2015','dd-mon-yyyy') FROM dual UNION ALL
    select 'end4',to_date('17-jan-2015','dd-mon-yyyy') from dual union all
    SELECT 'end5',to_date('19-jan-2015','dd-mon-yyyy') FROM dual UNION ALL
    SELECT 'end6',to_date('20-jan-2015','dd-mon-yyyy') FROM dual
  )
  ,start_table (start_id, start_dt) AS (
    select 'start0',to_date('31-dec-2014','dd-mon-yyyy') from dual union all
    select 'start1',to_date('03-jan-2015','dd-mon-yyyy') from dual union all
    select 'start2',to_date('05-jan-2015','dd-mon-yyyy') from dual union all
    select 'start3',to_date('07-jan-2015','dd-mon-yyyy') from dual union all
    select 'start4',to_date('08-jan-2015','dd-mon-yyyy') from dual union all
    SELECT 'start5',to_date('14-jan-2015','dd-mon-yyyy') FROM dual
  )
  , start_and_end as (
    select -1 rowtype, end_id id, end_dt dt from end_table
    union all
    select 1, s.* from start_table s
  )
  , running_count as (
    select se.*,
      sum(rowtype) over(order by dt, rowtype desc) cnt
    from start_and_end se
  )
  , filtered_ends as (
    select rowtype, id, dt from (
      select rc.*,
        min(decode(rowtype,-1,cnt)) over(
          order by dt, rowtype desc
          rows between unbounded preceding and 1 preceding
        ) mincnt
      from running_count rc
    )
    where cnt >= nvl(mincnt,0) or cnt >= 0
  )
  select * from (
    select rowtype, id,
    row_number() over(partition by rowtype order by dt) rn
    from filtered_ends
  )
  pivot(max(id) for rowtype in(1 st_id, -1 end_id))
  order by rn;
  

  ST_ID END_ID RN

  ---------- ------ ------

  1 start0 end0

  2 Start1 end2

  3 Start2 end3

  4 start3 bout4

  5 debut4 end5

  Debut5 6 end6

• Question of the Clause

  Hello I have a query like the following. ------------------------------------------------ with base as ( Select date, cust_id, amount several tables Union of all the Select date, cust_id, amount several tables ) Select * Basic where snapshot_date = July 31, 2014" ------------------------------------------------- This query is too long. If I throw ' where snapshot_date = July 31, 2014 ' ' for both selects in the "-with clause", it's really fast. This is because all the tables in the db have indices on the field of snapshot_date. It seems that the with clause creates a view with all stereotypes, rather than just July 31, 2014. Y at - there a way to force the "-with clause" to retrieve data for only snapshot_date = July 31, 2014 ' as specified in the main query above without using where clause within the "with clause"? Thanks a lot to all the...

  As Gaff has suggested, I begin to believe that the WITH Clause is actually a temporary display based only on what is covered by the clause.

  Maybe it did and maybe it isn't. There is NO value to speculate unless you use this speculation to actually test your assumptions.

  In my case, there are 30 years of history.

  So why do you need to do a UNION ALL operation to get the results? Why can't simply run both queries separately?

  The optimizer does not resemble what data are used.

  More speculation on your part. I can create a simple example of the SCOTT schema on 11.2.0.1.0 vanilla that uses the index.

  create the table emp_copy1 in select * from emp;

  create the table emp_copy2 in select * from emp;

  create index emp_copy1_ndx on emp_copy1 (deptno);

  create index emp_copy2_ndx on emp_copy2 (deptno);

  explain plan for
  with base as
  (
  Select ename, empno and deptno from emp_copy1
  Union of all the
  Select ename, empno and deptno from emp_copy2
  )
  Select * base where deptno = 20;

  Select * from table (dbms_xplan.display_cursor ())

  PLAN_TABLE_OUTPUT

  SQL_ID, 77q19ypwbjkvq, number of children 0

  -------------------------------------

  with base (select empno, ename, deptno emp_copy1 Union all the)

  Select empno, ename, deptno emp_copy2) select * base where

  DEPTNO = 20

  Hash value of plan: 2111487496

  -----------------------------------------------------------------------------------------------

  | ID | Operation | Name | Lines | Bytes | Cost (% CPU). Time |

  -----------------------------------------------------------------------------------------------

  |   0 | SELECT STATEMENT |               |       |       |     4 (100) |          |

  |   1.  VIEW                         |               |    10.   330.     4 (0) | 00:00:01 |

  |   2.   UNION-ALL |               |       |       |            |          |

  |   3.    TABLE ACCESS BY INDEX ROWID | EMP_COPY1 |     5.   165.     2 (0) | 00:00:01 |

  |*  4 |     INDEX RANGE SCAN | EMP_COPY1_NDX |     5.       |     1 (0) | 00:00:01 |

  |   5.    TABLE ACCESS BY INDEX ROWID | EMP_COPY2 |     5.   165.     2 (0) | 00:00:01 |

  |*  6 |     INDEX RANGE SCAN | EMP_COPY2_NDX |     5.       |     1 (0) | 00:00:01 |

  -----------------------------------------------------------------------------------------------

  Information of predicates (identified by the operation identity card):

  ---------------------------------------------------

  4 - access ("DEPTNO" = 20)

  6 - access ("DEPTNO" = 20)

  Note

  -----

  -dynamic sample used for this survey (level = 2)

  Which show the index range scan is done for both tables. These statistics are current since Oracle collected his stats when the indexes have been built

  I think that there is no solution to this problem

  . No, unless indicate you at least a way for us to reproduce the problem.

• Query to find data conneting that intersect each other.

  Hello

  I have a table like road_point_ids (script is below). Basically, this table have road_id with its start and its end point.

  I'm looking for a query that actually will connect each ID path beginning or end points.

  create the table road_point_ids

  (

  road_id number (10),

  start_point number (10),

  Number of end_point (10)

  );

  insert into road_point_ids

  values (1001, 10, 20);

  insert into road_point_ids

  values (1002, 20, 30);

  insert into road_point_ids

  values (1003, 30, 40);

  insert into road_point_ids

  values (1004, 40, 50);

  insert into road_point_ids

  values (1005, 50, 10);

  insert into road_point_ids

  values (1006, 70, 75);

  insert into road_point_ids

  values (1007, 30, 50);

  insert into road_point_ids

  values (1008, 10, 40);

  insert into road_point_ids

  values (1009, 80, 70);

  insert into road_point_ids

  values (1010, 90, 95);

  insert into road_point_ids

  values (1011, 50, 60);

  
  

  commit;

  SQL > select * from road_point_ids by 2;

  ROAD_ID START_POINT END_POINT

  -----------           -----------           -----------

  1001               10                    20

  1008               10                    40

  1002               20                    30

  1007               30                    50

  1003               30                    40

  1004               40                    50

  1011               50                    60

  1005               50                    10

  1006               70                    75

  1009               80                    70

  1010               90                    95

  11 selected lines

  In the example above only seven road_ids intersect at TWO points corresponding to start or end point.

  Output desired must be as...

  	ROAD_ID	START_POINT	END_POINT
  	1001	10	20
  	1002	20	30
  	1003	30	40
  	1004	40	50
  	1005	50	10
  	1007	30	50
  	1008	10	40

  7 selected lines

  I tried to use a lot of operator AND with the OR operator but sent to an incorrect result. Could someone provide me please help.

  Thanks in advance

  Saaz

  Hi, Saaz,

  How is this problem differs from your original problem?

  If it's just that the data is now contained in 2 tables (a few lines in a table, a few lines in the other) instead of a table, you can change my original solution by using a UNION to combine the two tables into one:

  WITH road_point_ids AS

  (

  SELECT road_id, start_point, end_point OF road_point_ids_1 UNION ALL

  SELECT road_id, start_point, end_point FROM road_point_ids_2

  )

  SELECT DISTINCT road_id, start_point, end_point

  OF road_point_ids

  WHERE CONNECT_BY_ISCYCLE = 1

  CONNECT BY NOCYCLE start_point = PRIOR end_point

  ;

  The WITH clause is new; the main request is that I posted earlier.

• Manual partitioning - with a check for pruning constraint

  I have a client who won't pay partitioning for real, but only Monster table on a data warehouse that must be broken into pieces for performance applications.

  What I want to do is manually partition table, create a union all join view which is located on top, then have oracle prune query plans for me to hit only the tables that I need to fill the result set.

  
  

  -Create table
  create the table a_tbl
  (
  col1 varchar2 (20).
  col2 varchar2 (20).
  COL3 varchar2 (20).
  COL4 varchar2 (20)
  )
  tablespace AAA_DATA;

  -Create/recreate check constraints
  ALTER table a_tbl
  Add constraint a_tbl_restrict
  check (col1 = 'aaa');
  
  create the table b_tbl
  (
  col1 varchar2 (20).
  col2 varchar2 (20).
  COL3 varchar2 (20).
  COL4 varchar2 (20)
  )
  tablespace AAA_DATA;

  -Create/recreate check constraints
  ALTER table b_tbl
  Add constraint b_tbl_restrict
  check (col1 = 'bbb');
  
  create the table c_tbl
  (
  col1 varchar2 (20).
  col2 varchar2 (20).
  COL3 varchar2 (20).
  COL4 varchar2 (20)
  )
  tablespace AAA_DATA;

  -Create/recreate check constraints
  ALTER table c_tbl
  Add constraint c_tbl_restrict
  check (col1 = 'ccc');
  
  ALTER table A_TBL
  Add primary key constraint a_tbl_pk (COL1, COL2)
  using index
  tablespace AAA_INDEX;

  ALTER table B_TBL
  Add primary key constraint b_tbl_pk (COL1, COL2)
  using index
  tablespace AAA_INDEX;

  ALTER table C_TBL
  Add primary key constraint c_tbl_pk (COL1, COL2)
  using index
  tablespace AAA_INDEX;
  
  Create view abc_v as
  Select * from a_tbl
  Union of all the
  Select * from b_tbl
  Union of all the
  Select * from c_tbl;

  When I run a query against the view that would eliminate two of the tables, I don't see the size that passes.

  Select * from abc_v

  where col1 = 'aaa' and col2 < '100'

  order by col2

  SELECT STATEMENT, GOAL = 3 1 48 ALL_ROWS

  SORT ORDER BY 3 1 48

  DISCOVERED A LOT OF 2-1-48 ABC_V

  UNION-ALL

  TABLE ACCESS BY INDEX ROWID BUNCH 4 2 96 A_TBL

  INDEX RANGE SCAN BUNCH 3 2 A_TBL_PK

  FILTER

  TABLE ACCESS BY INDEX ROWID HEAP 2 1 48 B_TBL

  INDEX RANGE SCAN HEAP 2 1 B_TBL_PK

  FILTER

  TABLE ACCESS BY INDEX ROWID HEAP 2 1 48 C_TBL

  INDEX RANGE SCAN HEAP 2 1 C_TBL_PK

  I do something that is not possible without partitioning "true"?  Or I put in place without a certain way?

  Thank you!

  Cory Aston

  Cory,

  the concept of the partition views is very old - among the best results in google, I get a link for the setting to version 7.3.3 guide: view of Partition Management - I didn't know that these pages still exist. 7.3 was released in 1996.

  Since then, Oracle has frequently said that views of partition are desupported - but they seem to work yet. If I use your example I get the following execution plan:

  --------------------------------------------------------------------------------------------

  | ID | Operation | Name | Lines | Bytes | Cost (% CPU). Time |

  --------------------------------------------------------------------------------------------

  |   0 | SELECT STATEMENT |          |     1.    48.     1 (100) | 00:00:01 |

  |   1.  SORT ORDER BY |          |     1.    48.     1 (100) | 00:00:01 |

  |   2.   VIEW                          | ABC_V |     1.    48.     0 (0) | 00:00:01 |

  |   3.    UNION-ALL |          |       |       |            |          |

  |   4.     TABLE ACCESS BY INDEX ROWID | A_TBL |     1.    48.     1 (0) | 00:00:01 |

  |*  5 |      INDEX RANGE SCAN | A_TBL_PK |     1.       |     1 (0) | 00:00:01 |

  |*  6 |     FILTER |          |       |       |            |          |

  |   7.      TABLE ACCESS BY INDEX ROWID | B_TBL |     1.    48.     1 (0) | 00:00:01 |

  |*  8 |       INDEX RANGE SCAN | B_TBL_PK |     1.       |     1 (0) | 00:00:01 |

  |*  9 |     FILTER |          |       |       |            |          |

  |  10.      TABLE ACCESS BY INDEX ROWID | C_TBL |     1.    48.     1 (0) | 00:00:01 |

  | * 11 |       INDEX RANGE SCAN | C_TBL_PK |     1.       |     1 (0) | 00:00:01 |

  --------------------------------------------------------------------------------------------

  Information of predicates (identified by the operation identity card):

  ---------------------------------------------------

  5 - access ("COL1" = 'aaa' AND "COL2"<>

  6 - filter (NULL IS NOT NULL)

  8 - access ("COL1" = 'aaa' AND "COL2"<>

  9 - filter (NULL IS NOT NULL)

  11 - access ("COL1" = 'aaa' AND "COL2"<>

  The most important part is (once again) the section predicate containing filter NULL IS NOT NULL predicates that should always evaluate to false. So I would say: works as advertised (once upon a time...)

  Concerning

  Martin

• Digital signage using SYS_CONTEXT

  Version 11.2.0.3

  I'm trying to create a view that refers to two identical paintings. The results will be presented only ever among the tables based on a variable SYS_CONTEXT.

  The issue I see is however that the implementation plan is different when I use literals as opposed to a variable SYS_CONTEXT. The plan with literals is what I expect to see when I use SYS_CONTEXT as well. The difference between the 2 plans, is that the filter filter (NULL IS NOT NULL) appears for the SELECT that does not run. The following illustrates the problem:

  create table tbl1 (id number, date_created date);

  create table tbl2 (id number, date_created date);

  insert into tbl1 values (1, to_date('20140114','yyyymmdd'));

  insert into tbl2 values (2, to_date('20140115','yyyymmdd'));

  insert into tbl2 values (2, to_date('20140115','yyyymmdd'));

  COMMIT;

  create or replace context tbl_switch using set_tbl_context;

  create or replace procedure Set_Tbl_Context (p_Tbl_No in varchar2) as
  Start
  DBMS_SESSION.set_Context ('TBL_SWITCH', 'SWITCH_VALUE', p_Tbl_No);
  end Set_Tbl_Context;
  /

  SQL > exec set_tbl_context ('2');

  PL/SQL procedure successfully completed.

  SQL > select sys_context ('TBL_SWITCH', 'SWITCH_VALUE') FROM DUAL;

  SYS_CONTEXT ('TBL_SWITCH', 'SWITCH_VALUE')
  ----------------------------------------
  2

  SQL > set autotrace on
  SQL > select * from tbl1 where 1 = 2
  2 Union all the
  3 select * from tbl2 where 2 = 2
  4.

  ID DATE_CREA
  ---------- ---------
  2 15 JANUARY 14
  2 15 JANUARY 14

  
  Execution plan
  --------------
  Hash value of plan: 6123662

  ----------------------------------------------------------------------------
  | ID | Operation | Name | Lines | Bytes | Cost (% CPU). Time |
  ----------------------------------------------------------------------------
  |   0 | SELECT STATEMENT |      |     3.    66.     3 (100) | 00:00:01 |
  |   1.  UNION-ALL |      |       |       |            |          |
  |*  2 |   FILTER |      |       |       |            |          |
  |   3.    TABLE ACCESS FULL | TBL1 |     1.    22.     3 (0) | 00:00:01 |
  |   4.   TABLE ACCESS FULL | TBL2 |     2.    44.     3 (0) | 00:00:01 |
  ----------------------------------------------------------------------------

  Information of predicates (identified by the operation identity card):
  ---------------------------------------------------

  2 - filter (NULL IS NOT NULL)

  Note
  -----
  -dynamic sample used for this survey (level = 2)

  
  Statistics
  ----------
  16 recursive calls
  0 db block Gets
  Gets 36 compatible
  0 physical reads
  0 redo size
  436 bytes sent via SQL * Net to client
  364 bytes received via SQL * Net from client
  2 SQL * Net back and forth to and from the client
  0 sorts (memory)
  0 sorts (disk)
  2 rows processed

  SQL > select * from tbl1
  where the 2 ' 1'= sys_context('TBL_SWITCH','SWITCH_VALUE')
  3 Union all the
  4 Select * from tbl2
  where the 5 ' 2'= sys_context('TBL_SWITCH','SWITCH_VALUE')
  6.

  ID DATE_CREA
  ---------- ---------
  2 15 JANUARY 14
  2 15 JANUARY 14

  
  Execution plan
  --------------
  Hash value of plan: 687287648

  ----------------------------------------------------------------------------
  | ID | Operation | Name | Lines | Bytes | Cost (% CPU). Time |
  ----------------------------------------------------------------------------
  |   0 | SELECT STATEMENT |      |     5:    66.     6 (50) | 00:00:01 |
  |   1.  UNION-ALL |      |       |       |            |          |
  |*  2 |   FILTER |      |       |       |            |          |
  |   5:    TABLE ACCESS FULL | TBL1 |     1.    22.     3 (0) | 00:00:01 |
  |*  4 |   FILTER |      |       |       |            |          |
  |   3:    TABLE ACCESS FULL | TBL2 |     2.    44.     3 (0) | 00:00:01 |
  ----------------------------------------------------------------------------

  Information of predicates (identified by the operation identity card):
  ---------------------------------------------------

  2 - filter (SYS_CONTEXT ('TBL_SWITCH', 'SWITCH_VALUE') = '1')
  4 - filter (SYS_CONTEXT ('TBL_SWITCH', 'SWITCH_VALUE') = '2')

  Note
  -----
  -dynamic sample used for this survey (level = 2)

  
  Statistics
  ----------
  7 recursive calls
  0 db block Gets
  Gets 24 consistent
  0 physical reads
  0 redo size
  436 bytes sent via SQL * Net to client
  364 bytes received via SQL * Net from client
  2 SQL * Net back and forth to and from the client
  0 sorts (memory)
  0 sorts (disk)
  2 rows processed

  Can anyone suggest why this is? In addition, the query with the variable SYS_CONTEXT does not two SELECT?

  Thank you

  That's exactly the plan that you want to see (although I must say I don't like the strategy that you are adopting).

  If the sys_context value is 1 then line 2 will call line 3 to run, but line 4 will be short circuit and do not call the line 5

  If the sys_content value is 2, then line 1 will be short-circuit and not dial 3, but call line 4 line 5

  Update: your method is similar to an example I talked some time ago:conditional SQL & #8211; 3. Notebook of the Oracle and something similar is generated internally by the particularly specific optimizer transformation: conditional SQL | Notebook of the Oracle

  It would not have been too difficult to set up a small test to see if this was true.

  Concerning

  Jonathan Lewis

• How to improve the performance of queries

  Hello everyone, this is the first time that I am responsible for investigating the performance of the queries.

  I have no permission to market the autotrace, so I just used the Plan to explain on the table on the development site.

  Here can someone tell me if there are obvious problems with the query below?

  This query is the basis for a single report. The amount of data on the Production site is about 25 times amount of data on the development site.

  User says that they must wait about 10 minutes for the report to load.

  Thank you very much in advance.

  Databases:

  Oracle 11 GR 1 matter (CARS)

  Application running the report is Apex Oracle version 4.0.2

  explain plan for

  Select vw. NPP, vw. PROFILE_NAME, vw. ADDRESS, vw. COMPETENCE, vw. ASSIGNED_CPO_ID,

  VW. STATUS_NAME, vw. DOMICILE_ID, vw. DOMICILE_NAME, vw. OVERPAYMENT_GROUP_ID, vw. CPO_NAME, vw. OCCURRENCES, Act. STATUS_CHANGE_DATE,

  (select name from opay_status where id = ACTION_ID) action_name,.

  ACTIONED_DATE, "CLOSED_BY,.

  (select name from OPAY_PAYMENT_SOURCE_TYPES where id = PAY_TYPE_ID) Payment_Source,

  VW. REVIEW_DATE

  ftrx.opay_overpayments_summary_vw vw, OPAY_OVERPAYMENT_ACTIONS Act

  where vw. OVERPAYMENT_GROUP_ID = Act.ID (+) and upper (Volkswagen. <>STATUS_NAME) "TRANSFORMED."

  and OPAY_PROCESSED_OVERPAYMENT (vw. OVERPAYMENT_GROUP_ID) = 'N'

  Union of all the

  Select RPC. NPP, CPP. PROFILE_NAME, CPP. ADDRESS, RPC. COMPETENCE, CPP. ASSIGNED_CPO_ID,

  CPP. STATUS_NAME, CPP. DOMICILE_ID, CPP. DOMICILE_NAME, CPP. OVERPAYMENT_GROUP_ID, CPP. ASSIGNED_CPO_NAME CPO_NAME, cpp.cycle_count OCCURRENCES, CPP. STATUS_CHANGE_DATE,

  CPP.action_name,

  CPP. ACTIONED_DATE, (select name of CPO where id = CPO_ID) CLOSED_BY,.

  CPP. Payment_Source,

  CPP. STATUS_CHANGE_DATE DUMMY_DATE

  the CPP OPAY_CPO_PROFILE_PROCESSED;

  /

  Select * from table (dbms_xplan.display);

  Hash value of plan: 3275359346

  -------------------------------------------------------------------------------------------------------------------------------

  | ID | Operation | Name                         | Lines | Bytes | TempSpc | Cost (% CPU). Time |

  -------------------------------------------------------------------------------------------------------------------------------

  |   0 | SELECT STATEMENT |                              |   260K |    65 M |       | 44949 (4) | 00:09:00 |

  |   1.  UNION-ALL |                              |       |       |       |            |          |

  |   2.   TABLE ACCESS BY INDEX ROWID | OPAY_STATUS |     1.    16.       |     1 (0) | 00:00:01 |

  |*  3 |    INDEX UNIQUE SCAN | OPAY_PK_STATUS |     1.       |       |     0 (0) | 00:00:01 |

  |   4.   TABLE ACCESS BY INDEX ROWID | OPAY_PAYMENT_SOURCE_TYPES |     1.    12.       |     1 (0) | 00:00:01 |

  |*  5 |    INDEX UNIQUE SCAN | OPAY_PK_SOURCE_TYPES |     1.       |       |     0 (0) | 00:00:01 |

  |*  6 |   EXTERNAL RIGHT HASH JOIN |                              |   197K |    86.       | 43459 (1) | 00:08:42 |

  |   7.    TABLE ACCESS FULL | OPAY_OVERPAYMENT_ACTIONS |  1224 | 29376 |       |   209 (0) | 00:00:03 |

  |   8.    VIEW                                | OPAY_OVERPAYMENTS_SUMMARY_VW |   197K |    51 M |       | 43249 (1) | 00:08:39 |

  |   9.     UNION-ALL |                              |       |       |       |            |          |

  | * 10 |      HASH JOIN |                              |     2.   570 |       | 17032 (1) | 00:03:25 |

  | * 11 |       VIEW                             |                              |     2.   560.       |  7456 (1) | 00:01:30 |

  |  12.        KIND OF WINDOW.                              |     2.   278.       |  7456 (1) | 00:01:30 |

  |  13.         NESTED EXTERNAL LOOPS |                              |     2.   278.       |  7455 (1) | 00:01:30 |

  |  14.          NESTED LOOPS |                              |     2.   242.       |  7453 (1) | 00:01:30 |

  | * 15 |           OUTER HASH JOIN |                              |     2.   214.       |  7451 (1) | 00:01:30 |

  |  16.            NESTED EXTERNAL LOOPS |                              |     2.   194.       |  7448 (1) | 00:01:30 |

  | * 17.             TABLE ACCESS FULL | OPAY_OVERPAYMENTS |     2.   168.       |  7447 (1) | 00:01:30 |

  |  18.             TABLE ACCESS BY INDEX ROWID | OPAY_OVERPAYMENT_ACTIONS |     1.    13.       |     1 (0) | 00:00:01 |

  | * 19.              INDEX UNIQUE SCAN | OPAY_PK_OVERPAY_ACTIONS |     1.       |       |     0 (0) | 00:00:01 |

  |  20.            TABLE ACCESS FULL | STATUS                       |     3.    30.       |     3 (0) | 00:00:01 |

  |  21.           TABLE ACCESS BY INDEX ROWID | HOME |     1.    14.       |     1 (0) | 00:00:01 |

  | * 22.            INDEX UNIQUE SCAN | PK_DOMICILE |     1.       |       |     0 (0) | 00:00:01 |

  |  23.          TABLE ACCESS BY INDEX ROWID | CPO                          |     1.    18.       |     1 (0) | 00:00:01 |

  | * 24.           INDEX UNIQUE SCAN | PK_CPO                       |     1.       |       |     0 (0) | 00:00:01 |

  |  25.       VIEW                             |                              |   197K |   963K |       |  9574 (1) | 00:01:55 |

  |  26.        HASH GROUP BY.                              |   197K |  7127K |  8544K |  9574 (1) | 00:01:55 |

  | * 27.         EXTERNAL RIGHT HASH JOIN |                              |   197K |  7127K |       |  7664 (1) | 00:01:32 |

  |  28.          TABLE ACCESS FULL | STATUS                       |     3.     9.       |     3 (0) | 00:00:01 |

  | * 29.          EXTERNAL RIGHT HASH JOIN |                              |   197K |  6549K |       |  7660 (1) | 00:01:32 |

  |  30.           TABLE ACCESS FULL | OPAY_OVERPAYMENT_ACTIONS |  1224 | 11016.       |   209 (0) | 00:00:03 |

  | * 31.           HASH JOIN |                              |   197K |  4816K |       |  7449 (1) | 00:01:30 |

  |  32.            INDEX SCAN FULL | PK_DOMICILE |    73.   292.       |     1 (0) | 00:00:01 |

  |  33.            TABLE ACCESS FULL | OPAY_OVERPAYMENTS |   197K |  4045K |       |  7446 (1) | 00:01:30 |

  | * 34 |      HASH JOIN |                              |   197K |    53 M |  3280K | 26217 (1) | 00:05:15 |

  |  35.       VIEW                             |                              |   197K |   963K |       |  9574 (1) | 00:01:55 |

  |  36.        HASH GROUP BY.                              |   197K |  7127K |  8544K |  9574 (1) | 00:01:55 |

  | * 37 |         EXTERNAL RIGHT HASH JOIN |                              |   197K |  7127K |       |  7664 (1) | 00:01:32 |

  |  38.          TABLE ACCESS FULL | STATUS                       |     3.     9.       |     3 (0) | 00:00:01 |

  | * 39 |          EXTERNAL RIGHT HASH JOIN |                              |   197K |  6549K |       |  7660 (1) | 00:01:32 |

  |  40.           TABLE ACCESS FULL | OPAY_OVERPAYMENT_ACTIONS |  1224 | 11016.       |   209 (0) | 00:00:03 |

  | * 41.           HASH JOIN |                              |   197K |  4816K |       |  7449 (1) | 00:01:30 |

  |  42.            INDEX SCAN FULL | PK_DOMICILE |    73.   292.       |     1 (0) | 00:00:01 |

  |  43.            TABLE ACCESS FULL | OPAY_OVERPAYMENTS |   197K |  4045K |       |  7446 (1) | 00:01:30 |

  | * 44 |       VIEW                             |                              |   197K |    52 M |       | 13757 (1) | 00:02:46 |

  |  45.        KIND OF WINDOW.                              |   197K |    26 M |    28 M | 13757 (1) | 00:02:46 |

  | * 46 |         EXTERNAL RIGHT HASH JOIN |                              |   197K |    26 M |       |  7673 (1) | 00:01:33 |

  |  47.          TABLE ACCESS FULL | CPO                          |    91.  1638.       |     4 (0) | 00:00:01 |

  | * 48 |          EXTERNAL RIGHT HASH JOIN |                              |   197K |    22 M |       |  7667 (1) | 00:01:33 |

  |  49.           TABLE ACCESS FULL | STATUS                       |     3.    30.       |     3 (0) | 00:00:01 |

  | * 50 |           EXTERNAL RIGHT HASH JOIN |                              |   197K |    20 M |       |  7663 (1) | 00:01:32 |

  |  51.            TABLE ACCESS FULL | OPAY_OVERPAYMENT_ACTIONS |  1224 | 15912 |       |   209 (0) | 00:00:03 |

  | * 52 |            HASH JOIN |                              |   197K |    18 M |       |  7452 (1) | 00:01:30 |

  |  53.             TABLE ACCESS FULL | HOME |    73.  1022 |       |     3 (0) | 00:00:01 |

  | * 54 |             TABLE ACCESS FULL | OPAY_OVERPAYMENTS |   197K |    15 M |       |  7448 (1) | 00:01:30 |

  |  55.   TABLE ACCESS BY INDEX ROWID | CPO                          |     1.    18.       |     1 (0) | 00:00:01 |

  | * 56 |    INDEX UNIQUE SCAN | PK_CPO                       |     1.       |       |     0 (0) | 00:00:01 |

  |  57.   TABLE ACCESS FULL | OPAY_CPO_PROFILE_PROCESSED | 63108 |  9429K |       |  1490 (1) | 00:00:18 |

  -------------------------------------------------------------------------------------------------------------------------------

  Information of predicates (identified by the operation identity card):

  ---------------------------------------------------

  3 - access("ID"=:B1)

  5 - access("ID"=:B1)

  6 - access("VW".") OVERPAYMENT_GROUP_ID "=" LAW ". "ID" (+)) "

  10 - access("A".") WITH THE ID '=' B '. (' ' ID ')

  11 filter (SUPERIOR ("a.") "<>STATUS_NAME"), "TRANSFORMED" AND "OPAY_PROCESSED_OVERPAYMENT"("A"."" OVERPAYMENT_GROUP_ID') = 'N')

  15 - access("STS".") ID "(+) = TO_NUMBER ("ACT"". ")" STATUS'))

  17 - filter("OPAY".") (OVERPAYMENT_GROUP_ID"IS NULL)

  19 - access("OPAY".") OVERPAYMENT_GROUP_ID "=" LAW ". "ID" (+)) "

  22 - access("OPAY".") DOMICILE_ID "=" DOM ". (' ' ID ')

  24 - access("ACT".") ASSIGNED_CPO_ID "=" DPC ". "ID" (+)) "

  27 - access("STS".") ID "(+) = TO_NUMBER ("ACT"". ")" STATUS'))

  29 - access("OPAY".") OVERPAYMENT_GROUP_ID "=" LAW ". "ID" (+)) "

  31 - access("OPAY".") DOMICILE_ID "=" DOM ". (' ' ID ')

  34 - access("A".") WITH THE ID '=' B '. (' ' ID ')

  37 - access("STS".") ID "(+) = TO_NUMBER ("ACT"". ")" STATUS'))

  39 - access("OPAY".") OVERPAYMENT_GROUP_ID "=" LAW ". "ID" (+)) "

  41 - access("OPAY".") DOMICILE_ID "=" DOM ". (' ' ID ')

  44 filter (SUPERIOR ("a.") "<>STATUS_NAME"), "TRANSFORMED" AND "OPAY_PROCESSED_OVERPAYMENT"("A"."" OVERPAYMENT_GROUP_ID') = 'N')

  46 - access("ACT".") ASSIGNED_CPO_ID "=" DPC ". "ID" (+)) "

  48 - access("STS".") ID "(+) = TO_NUMBER ("ACT"". ")" STATUS'))

  50 - access("OPAY".") OVERPAYMENT_GROUP_ID "=" LAW ". "ID" (+)) "

  52 - access("OPAY".") DOMICILE_ID "=" DOM ". (' ' ID ')

  54 - filter("OPAY".") OVERPAYMENT_GROUP_ID' IS NOT NULL)

  56 - access("ID"=:B1)

  - And this is the content of the function used in the query

  create or replace

  FUNCTION "opay_PROCESSED_OVERPAYMENT" (NUMBER p_opay_group_id) RETURN varchar AS

  fake number;

  result varchar (1);

  BEGIN

  If p_opay_group_id is null returns 'n';

  on the other

  SELECT count (distinct is_processed_yn) in a model of opay_overpayments where overpayment_group_id = p_opay_group_id;

  If dummy <>1 then back 'n';

  On the other

  Select distinct is_processed_yn as a result of opay_overpayments where overpayment_group_id = p_opay_group_id;

  return the result;

  end if;

  end if;

  END;

  - And here are the details of the view.

  CREATE OR REPLACE VIEW 'OPAY_OVERPAYMENTS_SUMMARY_VW '.

  As

  Select one. «' REVIEW_DATE ', a. "ID", a. "NPP", a. "PROFILE_NAME «, a» ADDRESS", a. "COMPETENCE", a"ASSIGNED_CPO_ID", a. "STATUS_NAME", a. "DOMICILE_ID", a. "DOMICILE_NAME", a. "OVERPAYMENT_GROUP_ID", a. "CPO_NAME", a. "OCCURRENCES»

  (select review_date, id, NPP, profile_name, address

  domicile_name, status_name, assigned_cpo_id, competence

  overpayment_group_id, cpo_name

  , count (NPP) over (PARTITION BY ppn) AS OCCURRENCES

  (SELECT review_date, id, NPP, profile_name, address, assigned_cpo_id

  status_name, domicile_id, domicile_name

  overpayment_group_id, cpo_name

  OF ftrx.opay_overpayments_vw

  where overpayment_group_id is null)) a

  Join (select max (id) id, overpayment_group_id, NPP group opay_overpayments_vw by overpayment_group_id, ppn) b

  on all the a.id = b.id union

  Select one. "' REVIEW_DATE ', a." ID ", a." NPP ", a." PROFILE_NAME «, a» ADDRESS ", a." COMPETENCE ", a" ASSIGNED_CPO_ID ", a." STATUS_NAME ", a." DOMICILE_ID ", a." DOMICILE_NAME ", a." OVERPAYMENT_GROUP_ID ", a." CPO_NAME ", a." OCCURRENCES "(select review_date, id, NPP

  profile_name

  address

  skill

  assigned_cpo_id

  status_name

  domicile_id

  domicile_name

  overpayment_group_id

  cpo_name

  , count (overpayment_group_id) over (PARTITION BY overpayment_group_id) AS OCCURRENCES

  (SELECT review_date, id, NPP

  profile_name

  address

  skill

  assigned_cpo_id

  status_name

  domicile_id

  domicile_name

  overpayment_group_id

  cpo_name

  OF ftrx.opay_overpayments_vw

  where overpayment_group_id is not null)) a

  Join (select max (id) id, overpayment_group_id, NPP group opay_overpayments_vw by overpayment_group_id, ppn) b

  on a.id = b.id

  Let me know if my explanation is not clear enough.

  Thanks for reading.

  I thought that the function can be part of the problem. He called lots and made several requests by rank. Calls to PL/SQL in SQL a) require a context switch in the execution of the query (from SQL, PL/SQL) and (b) are completely opaque to the optimizer - it can not re - write to be faster than the procedure one call at a time. Instead of trying to re - write that it is faster, try to write it out of the query completely by turning it into a piece of SQL in the container view.

  Assuming that opay_overpayments.is_processed_yn is OPAY_PROCESSED_OVERPAYMENT(vw., changement ou «N», «Y» OVERPAYMENT_GROUP_ID) = 'N'

  in

  AND (vw. OVERPAYMENT_GROUP_ID or vw. OVERPAYMENT_GROUP_ID in

  (select OVERPAYMENT_GROUP_ID in the opay_overpayments where is_processed_yn = ' n and OVERPAYMENT_GROUP_ID is not null)

  )

  Having done this, you can probably also turn the query into an outer join between vw, the OPAY_OVERPAYMENT_ACTIONS Act ftrx.opay_overpayments_summary_vw and (select OVERPAYMENT_GROUP_ID in the opay_overpayments where is_processed_yn = ' only and OVERPAYMENT_GROUP_ID is not null) if the optimizer which has not already done so. That would minimize the amount of querying.

• NOT IN and ANTI JOIN

  Hello guys,.
  
  with the example I want to insert the new translation into a table of conversion below, but
  the NOT IN clause does not work. I also tried an ANTI-JOINTURE after reading some
  why NOT IN may not work but the ANTI-JOINTURE did not work either.
  The result of the query above is 0 rows inserted. There are about 1000 lines that already exist and I want to avoid
  insertion of those, once again. Note that the text is not even the ID or the SOURCE.
  
  INSERT INTO T_Translations
  (
  SELECT ID, SOURCE, TEXT, OF
  (
  SELECT distinct ID, SOURCE, DOMAIN,
  TransText AS TEXT
  DE)
  
  -Deutsche text aus TKZ_1
  SELECT TransText OF
  -... - Source table 1
  UNION ALL
  -... - Source table 2
  UNION ALL
  -... - Source table 3
  UNION ALL
  -... - Source table 4
  
  ) WHERE TransText IS NOT NULL
  )
  
  WHEN NOT IN TEXT (SELECT TEXT FROM T_Translations)
  );
  
  Any suggestions?
  
  Thank you very much.

  It might be a problem with the alias.

  The TEXT column can come from many different parts of your query. I changed it a bit and removed unnecessary inline views.

  INSERT INTO T_Translations i
    SELECT distinct v.ID, v.SOURCE, v.DOMAIN, v.TransText
    FROM (
                --Deutsche Texte aus TKZ_1
                SELECT TransText FROM
                --...--Source Table 1
                UNION ALL
                --...--Source Table 2
                UNION ALL
                --...--Source Table 3
                UNION ALL
                --...--Source Table 4
                ) v
     WHERE v.TransText IS NOT NULL
     AND not exists (SELECT null FROM T_Translations t2 where v.TransText = t2.TEXT )
    ;
  

• Character of sql, simple sorting

  Hello
  
  I have three SQLS which I'm running with 'Union '.
  
  for example
  
  Select col1, count (*) in the table where type = 'pieces '.
  Group of col1
  Union
  Select col1, count (*) in the table where type = "consumables".
  Group of col1
  Union
  Select 'Total', count (*) table
  Group by 'Total '.
  
  I want to force the total column or the last row to select it and the column of pieces to always be the first line. Any suggestions? I tried something like prefixing the col1 with an alphabet. But that appears in the output also. Was looking for a better way!
  
  Thank you
  Sun

  Select col1, cnt,
  from (select 1 order2, col1, count (*) NTC table where type = 'parts')
  Group of col1
  Union
  Select 2, col1, count (*) from table where type = "consumables".
  Group of col1
  Union
  Select 3, 'Total', count (*) table
  control group by 'Total') by order2.