Clarification on NULL or behavior LEFT JOIN behavior

Similar Questions

• Interesting behavior of left join

  We will create 3 tables as follows:
  

  
  

  CREATE TABLE "A1" ()"A1_ID" NUMBER ENABLE NON NULL,CONSTRAINT "A1_PK" SELECT the PRIMARY KEY ('A1_ID') );

  CREATE TABLE "A2" ()"A2_ID" NUMBER ENABLE NON NULL,CONSTRAINT "A2_PK" SELECT the PRIMARY KEY ('A2_ID'));

  CREATE TABLE "A12" ()'A12_ID' ENABLE NUMBER NOT NULL,'A1_ID' ENABLE NUMBER NOT NULL,'A2_ID' ENABLE NUMBER NOT NULL,"A12_PK" PRIMARY KEY ('A12_ID'), CONSTRAINTREFERENCES of FOREIGN ("A1_ID") of CONSTRAINT "A12_A1_FK1" "BUTUNLESIK" KEY "" A1 "("A1_ID") ENABLE,KEY REFERENCES FOREIGN ("A2_ID") of CONSTRAINT 'A12_A2_FK1' 'BUTUNLESIK '. ("" A2 "("A2_ID") ALLOW);

  We insert the values as follows:

  Insert into A1 values (1);

  Insert into A1 values (2);

  Insert into A1 values (3);

  Insert 4 A1;

  Insert into A1 values (5);

  Insert into A1 values (6);

  Insert into A2 values (1);

  insert into values A12 (1,1,1);

  Tables a1 and A2 are just simple tables with primary keys and A12 table contains many-to-many relationship between them.

  When I run the query below,

  SELECT * A1 A1 A12 A12 LEFT JOIN ON a1.a1_id = a12.a1_id A2 LEFT JOIN A2 ON a2.a2_id = a12.a2_id WHERE a2.a2_id IS NULL;

  I get the following results, as I predicted:

  A1_ID A12_ID A1_ID A2_ID A2_ID

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

  6

  5

  4

  3

  2

  However when I run the second query below (where I select only the a1 id column instead of "*")

  SELECT a1.a1_id A1 A1 A12 A12 LEFT JOIN ON a1.a1_id = a12.a1_id A2 LEFT JOIN A2 ON a2.a2_id = a12.a2_id WHERE a2.a2_id IS NULL;

  
  

  I get nothing, no lines have been selected. When I look at the plan of the second query I see a predicate to filter such as "'NULL IS NOT NULL". " I guess that this does not return results. Is this a bug, or y at - it a satisfactory explanation for this behavior?

  Step 1 - check the execution plans.

  At first glance, this looks like a variant of https://forums.oracle.com/thread/2594321

  Concerning

  Jonathan Lewis

• LEFT JOIN and performance

  Hi guys,.

  I inherited a previous devloper the query below. T_MAXVALUES has 100,000 rows and T_STA_ERG about 1 m.

  The performance is so bad, it works for 5 hours now and I have nothing so I can´t say much about the data of test or how it should look.

  It works on Oracle 11 g and I can´t get my head around all JOIN clauses and the ultimate goal but sense that there could be a better way

  to achieve what is tried below. All of you guru´s able to immediately spot the big failures?

  SELECT DISTINCT T1. PARAMETER, T1. ZERL_MATRIX, T1. MONTHS, T1. ADDITIONAL_CRI,

  DECODE (H2. MRL, NULL, (DECODE (H1. MRL, NULL, (DECODE (H3. MRL, NULL, (DECODE (H4. MRL, NULL, NULL, 'R')), 'R')), M1. MRL)), H2. MRL)

  OF T_STA_ERG T1

  LEFT JOIN (SELECT PARAMETER, ZERL_MATRIX_2, MRL

  OF T_MAXVALUES, WHERE ZERL_MATRIX_1 IS NULL) H1

  ON T1. PARAMETER = H1. PARAMETER

  AND substr (T1. ZERL_MATRIX, 4, 3) = H1. ZERL_MATRIX_2

  LEFT JOIN (SELECT PARAMETER, ZERL_MATRIX_1, ZERL_MATRIX_2, MRL

  OF T_MAXVALUES

  WHERE ZERL_MATRIX_1 IS NOT NULL) H2

  ON T1. PARAMETER = H2. PARAMETER

  AND substr (T1. ZERL_MATRIX, 1, 3) = H2. ZERL_MATRIX_1

  AND substr (T1. ZERL_MATRIX, 4, 3) = H2. ZERL_MATRIX_2

  LEFT JOIN (SELECT PARAMETER, ZERL_MATRIX_2, MRL

  OF T_MAXVALUES

  WHERE ZERL_MATRIX_1 IS NULL) H3

  ON T1. PARAMETER = H3. PARAMETER

  LEFT JOIN T_MAXVALUES H4

  ON substr (T1. ZERL_MATRIX, 1, 3) = H4. ZERL_MATRIX_1

  The idea is to get maxvalues lines that at least match the lines of t. The game minimum is the game of columns of the parameter (type 3), or matches the matrix 1 (type 4).

  If you get all possible matches, and you will then take the MRLS of the best match: better if the parameter, matrix1 and matrix2 all game, then if parameter and matrix2 match, etc.

  What I have not included here are a test for null matrix1. If a match where parameter and matrix2 match, but there's a matrix1 value that does not match the value in T matrix1, will have to be thrown out, then just add this additional condition in the CASE statement.

  Untested of course:

  Select T1. PARAMETER, T1. ZERL_MATRIX, T1. MONTHS, T1. ADDITIONAL_CRI, Max (MRL) keep dense_rank of first order by type)

  Of

  (SELECT T1. PARAMETER, T1. ZERL_MATRIX, T1. MONTHS, T1. ADDITIONAL_CRI,

  -case when T1. PARAMETER = H.PARAMETER and substr (T1. ZERL_MATRIX, 1, 3) = H2. ZERL_MATRIX_1 AND substr (T1. ZERL_MATRIX, 4, 3) = H2. ZERL_MATRIX_2 then 1

  When T1. PARAMETER = H.PARAMETER and substr (T1. ZERL_MATRIX, 4, 3) = H1. ZERL_MATRIX_2 then 2

  When T1. PARAMETER = H.PARAMETER then 3

  When substr (T1. ZERL_MATRIX, 1, 3) = H.ZERL_MATRIX_1 then 4

  end match_type, h.mrl

  OF T_STA_ERG T1

  LEFT JOIN T_MAXVALUES h on (T1. PARAMETER = H.PARAMETER or substr (T1. (ZERL_MATRIX, 1, 3) = H.ZERL_MATRIX_1)

  )

  Group T1. PARAMETER, T1. ZERL_MATRIX, T1. MONTHS, T1. ADDITIONAL_CRI;

• LEFT JOIN increases the number of lines

  Hi guys,.

  I had a problem, my left join retrieves multiple values. I know he has only 252 in there that correspond to the place where

  condition. If I use the table in a left join with the same condition where my row count increases.

  -1176 lines

  Select count (erg_ID) of

  MySchema. T_STA_ERG sta_erg

  INNER JOIN T_MEN hoechst

  ON sta_erg. PARAMETER = hoechst. PARAMETER

  AND sta_erg. JAHR = 2014

  where sta_erg. MESSERG_KNG = 'A' AND sta_erg. MESSERG_ALPHA IN ('03 ") and sta_erg. NORM_MESS is null

  -252 lines

  Select distinct erg_ID myschema. T_STA_ERG sta_erg where sta_erg. MESSERG_KNG = 'A' AND sta_erg. MESSERG_ALPHA IN ('03 ") and sta_erg. NORM_MESS is null

  any clue´s how I can build in conditions in my join which would not increase the results of the line?

  Why not just an inner join then?

• Double left join creates a unwanted loop...

  I have therefore two paintings, I need to extract the counts of "tickets".

  I want to have the date given match then I use a 'connect by level' so that the same dates without data will have records in the result set.

  There are two tables I want to join on this "timeline" when I join each of them that individually, it works but when I join together that there is an order of operations issue I'm not sure how to move.

  ORIGINAL DATA:

  TABLE SMINCREQ:

  OPEN_DATE	NUMBER
  JANUARY 12, 2015	IM392039
  JANUARY 12, 2015	IM399495
  JANUARY 12, 2015	RM394950
  JANUARY 13, 2015	IM394958
  JANUARY 13, 2015	RM394958

  TABLE SMINTERACTIONS:

  OPEN_DATE	INCIDENT_ID
  JANUARY 12, 2015	SD394858
  JANUARY 12, 2015	SD399495
  JANUARY 12, 2015	SD394950
  JANUARY 13, 2015	SD394958
  JANUARY 13, 2015	SD394958

  THE QUERY:

  SELECT to_char(DAYS_,'DD-MON-YYYY'),
        
        sum(decode(INSTR(SMINCREQ."NUMBER",'IM'),'1','1',0)) "INCIDENT",
        sum(decode(INSTR(SMINCREQ."NUMBER",'RM'),'1','1',0)) "REQUEST"
       --sum(decode(INSTR(SMINTERACTIONS.INCIDENT_ID,'SD'),'1','1',0)) "INTERACTION"  
        FROM
    ( SELECT (TRUNC(to_date(SYSDATE-ROWNUM),'DD')) DAYS_ FROM DUAL
  CONNECT BY LEVEL <= (SELECT 14 FROM DUAL)
    ) THE_TIMELINE
      LEFT OUTER JOIN SMINCREQ ON  (to_char(DAYS_,'DD-MON-YYYY')=to_char(SMINCREQ.OPEN_TIME,'DD-MON-YYYY')  and (OPEN_GROUP like '%HELP%'))
      --LEFT OUTER JOIN SMINTERACTIONS on (to_char(DAYS_,'DD-MON-YYYY')=to_char(SMINTERACTIONS.OPEN_TIME,'DD-MON-YYYY')  and PRIMARY_ASSIGNMENT_GROUP like '%HELP%')
      GROUP BY to_char(DAYS_,'DD-MON-YYYY') ORDER BY to_char(DAYS_,'DD-MON-YYYY')
    ;
  
  

  When I run the present and just get the number of SMINCREQ is

  DATE	INCIDENTS	REQUEST
  JANUARY 12, 2015	103	4
  JANUARY 13, 2015	59	10

  When I run the query for SMINTERACTIONS I get

  DATE	INTERACTION
  JANUARY 12, 2015	55
  JANUARY 13, 2015	50

  When I try to run the two left joins them together I get:

  DATE	INCIDENTS	APPLICATIONS	INTERACTIONS
  JANUARY 12, 2015	5665	220	5885
  JANUARY 13, 2015	2950	500	3450

  I know that what is happening is a loop where the date is is developed in a relationship 1 to many through the joints and then the GROUP is not the case until the end.

  What I want to achieve is:

  DATE	INCIDENTS	REQIEST	INTERACTION
  JANUARY 12, 2015	103	4	55
  JANUARY 13, 2015	59	10	50

  Thanks for the tips

  I decided to go this route that will just join the new data (INTERACTIONS) of INCIDENTS already implemented and APPLICATIONS:

  SELECT NULL LINK,DAYS,INCIDENT,"REQUEST",sum(decode(INSTR(SMINTERACTIONS.INCIDENT_ID,'SD'),'1','1',0)) "INTERACTION" from(
  
  SELECT to_char(DAYS_,'DD-MON-YYYY') DAYS,sum(decode(INSTR(SMINCREQ."NUMBER",'IM'),'1','1',0)) "INCIDENT",
        sum(decode(INSTR(SMINCREQ."NUMBER",'RM'),'1','1',0)) "REQUEST"
        FROM
    ( SELECT (TRUNC(to_date(SYSDATE-ROWNUM),'DD')) DAYS_ FROM DUAL
   CONNECT BY LEVEL <= (SELECT 14 FROM DUAL)
    ) THE_TIMELINE
      LEFT OUTER JOIN SMINCREQ ON  (to_char(DAYS_,'DD-MON-YYYY')=to_char(SMINCREQ.OPEN_TIME,'DD-MON-YYYY')  and (OPEN_GROUP like '%HELP CENTER%')
  )
  
      GROUP BY to_char(DAYS_,'DD-MON-YYYY') ORDER BY to_char(DAYS_,'DD-MON-YYYY')
  ) LEFT JOIN SMINTERACTIONS on (DAYS=to_char(OPEN_TIME,'DD-MON-YYYY') and (PRIMARY_ASSIGNMENT_GROUP like '%HELP CENTER%'))
  GROUP BY DAYS,INCIDENT,"REQUEST" order by DAYS ASC;
  

  Thanks for all the entries...

• Left join

  Hello world

  Im trying to create the query that displays the values that are not available in the other table.

  Scenario:

  I have two tables, the material and materialfolder. I want to display documents that are not available in materialfolder.

  Here's my query

  SELECT m.material_id
      ,  m.MATERIAL_TYPE_ID
      ,  m.name
      ,  m.created
          FROM material m
          LEFT  JOIN materialfolder mf ON m.material_id = mf.material_id
          where to_char(m.CREATED, 'yyyy') >= ('2011')
          and to_char(m.CREATED, 'yyyy') <= ('2013');
  
  

  The result does not show the correct values. Don't know what my error here...

  Hoping that someone could help me.

  Thank you.

  Kind regards

  Ed

  Hello

  If you just want to know the values that are not available in materialfolder, use a NOT EXISTS condition:

  SELECT m.material_id

  m.MATERIAL_TYPE_ID

  $m.name

  m.created

  MATERIAL m

  If NOT EXISTS (select null from materialfolder mf

  where m.material_id = mf.material_id

  )

  AND m.CREATED > = TO_DATE('01/01/2011', 'dd/mm/yyyy') AND m.CREATED< to_date('01/01/2014',="">

• How to avoid the union and use a left join

  Hello

  I have the following tables

  {code}

  create the table mainTable

  (col1 varchar2 (2) varchar2 (2) col2, col3 varchar2 (2), col4 varchar2 (2));

  create table secTable

  (col1 varchar2 (2));

  create the table secTable1

  (col2 varchar2 (2));

  create the table secTable2

  (col3 varchar2 (2));

  Insert in maintable

  values ('a', 'b', 'this,' would be ');

  Insert in maintable

  values (', 'o', 'n', 'p');

  insert into secTable

  values ('a');

  insert into secTable1

  values ('b');

  insert into secTable2

  values ('c');

  commit;

  {code}

  I made a request, drank, I would use join without clause union how can I change to receive the same result as the following query?

  {code}

  	Select t1.*, 'bad value' | T1.Col1
  	from maintable t1
  	T1.Col1 left join t2 = t2.col1 sectable
  	where t2.col1 is null
  	Union of all the
  	Select t1.*, 'bad value' | T1.col2
  	from maintable t1
  	Join t2 left on t1.col1 = t2.col2 sectable1
  	where t2.col2 is null
  	Union of all the
  	Select t1.*, 'bad value' | T1. COL3
  	from maintable t1
  	Join t2 left on t1.col1 = t2.col3 sectable2
  	where t2.col3 is null

  {code}

  Thank you

  Coco

  Like this?:

  SQL > SELECT col1, col2, col3, col4, xvalue

  2 from (SELECT m0.*

  3, NVL (t0.col1, 'bad value t0' | m0.col1) t0_col1

  4, NVL (t1.col2, 'bad value t1' | m0.col2) t0_col2

  5, NVL (t2.col3, 'bad value t2' | m0.col3) t0_col3

  Maintable m0 6

  LEFT OUTER JOIN sectable t0 7

  8. WE (m0.col1 = t0.col1

  9 AND t0.col1 IS NULL)

  10 LEFT OUTER JOIN sectable1 t1

  11. WE (m0.col1 t1.col2 =

  12 AND t1.col2 IS NULL)

  13 LEFT OUTER JOIN t2 sectable2

  14. WE (m0.col1 = t2.col3

  15 t2.col3 AND IS NULL))

  16 UNPIVOT INCLUDES NULL values (xvalue wrongly IN (t0_col1, t0_col2, t0_col3))

  17 ORDER BY xvalue

  18.

  COL1 COL2 COL3 COL4 XVALUE

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

  a b c d bad value t0 a

  m o n p bad value m t0

  a b c d bad value t1 b

  m o n p o t1 value of evil

  a b c d bad value t2 c

  m o n p value t2 n bad

  6 selected lines.

• Left join of the two tables and multiple values into a single value separated by commas

  Hello

  I have following tables with their structures and their data as below.

  CREATE TABLE 'BETODI '. "" BETINFO ".

  (

  VARCHAR2 (8 BYTE) "CURRENTPRESS."

  ENABLE 'TYPEIDCONTAINER' VARCHAR2 (30 BYTE) NOT NULL

  )

  INSERT INTO Betinfo (Currentpress, typeidcontainer) VALUES ('A24G', 'PMC');

  INSERT INTO Betinfo (Currentpress, typeidcontainer) VALUES ('A24D', 'Pensky-MARTENS');

  INSERT INTO Betinfo (Currentpress, typeidcontainer) VALUES ("A25D", "CMP");

  INSERT INTO Betinfo (Currentpress, typeidcontainer) VALUES ('A25G', 'PMC');

  INSERT INTO Betinfo (Currentpress, typeidcontainer) VALUES ('A26D', 'PMC');

  INSERT INTO Betinfo (Currentpress, typeidcontainer) VALUES ('A26G', 'PMC');

  INSERT INTO Betinfo (Currentpress, typeidcontainer) VALUES ("A32G", "V-BFC3");

  INSERT INTO Betinfo (Currentpress, typeidcontainer) VALUES ('A32D', "V-BFC2");

  CREATE TABLE 'BETODI '. "" BETMASTER ".

  (

  ACTIVATE THE "CUREPRESS" TANK (5 BYTES) NOT NULL,

  ACTIVATE THE "TYPE" VARCHAR2 (5 BYTE) NOT NULL,

  NUMBER (5.0) "LASTPCIRIM".

  )

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ('A24', '45 M 8', 15);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ('A25', 42 16', 15);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ("A26", 16' 45, 15);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ("A27", '45 M 34', 16);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ('A28', '45 M 34', 16);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ('A29', '45 M 34', 16);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ('A30', '45MCH', 15);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ("A31", "45MCH", 16);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ('A32', '45MCH', 16);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ('A33', '45MCH', 16);

  INSERT INTO BetMaster (Curepress, type, lastpcirim) VALUES ("A34", "45MCH", 16);

  These two tables have left join as

  BETMASTER. CUREPRESS = substr (BETINFO. CURRENTPRESS, 1, 3)

  now I want to have the data in the two tables with fields Curepress, Lastpcirim, typeidcontainer.

  Also something like

  Make a group of typeidcontainer if this value is greater than 1 by press separated the values of semicolon (;)

  So, for example above, we should be given as

  A24 PMC 15; PENSKY-MARTENS

  A25 15 PMC

  A26 15 PMC

  A27 16 (NULL)

  A28 16 (NULL)

  A30 15 (NULL)

  A31 16 (NULL)

  A32 16 BFC2-V; V BFC3

  A33 16 (NULL)

  A34 16 (NULL)

  How could do?

  My current request is as

  Select distinct Curepress, lastpcirim, typeidcontainer

  BETMASTER STD left join INF BETINFO

  on the trim (STD. CUREPRESS) = substr (trim (INF. CURRENTPRESS), 1, 3)

  but I am unable to get the values separated by commas.

  Any help would be appreciated.

  Thank you

  Mahesh.

  Hi, Mahesh,

  If you want to only 1 row of output for each distinct combination of currentpress and lastpcirim?  This sounds like a job for GROUP BY.

  And you want the row to contain a list of all different typidcontainers-delimited?  This sounds like a job for the aggregate LISTAGG function.

  WITH joined_data AS

  (

  SELECT DISTINCT

  MST.curepress, mst.lastpcirim, inf.typeidcontainer

  OF betmaster STD

  LEFT JOIN betinfo ON TRIM (mst.curepress) inf = SUBSTR (TRIM (inf.currentpress)

  1

  3

  )

  )

  SELECT curepress, lastpcirim

  LISTAGG (typeidcontainer, ',')

  THE Group (ORDER BY typeidcontainer) AS container_list

  OF joined_data

  Curepress GROUP, lastpcirim

  ;

  Unfortunately, you can't say LISTAGG (DISTINCT ...), so you should always get the separate containers how you already are.  (Note that the subquery is just what you posted).

  Thanks for posting the CREATE TABLE and INSERT statements; It is very useful.  Don't forget to tell what version of Oracle you are using.  LISTAGG was new in Oracle 11.2.

  Why not add CHECK constraints (and perhaps triggers) to your tables, so that curepress and currentpress are not stored with the head or trailing spaces?  Then you wouldn't need to use the PAD in queries like this, and your code would be simpler and more effective.

• Left join between two tables using two different conditions

  I have following three tables with their data, as shown below.
  
  CREATE TABLE TIREMASTERPROCESS_TEMP
  (
  PRODUCTIONCODE TANK (12 BYTES),
  FIELDNAME CHAR (12 BYTES),
  DATACHR VARCHAR2 (60 BYTE),
  REVISIONNO TANK (3 BYTES),
  DATANUM NUMBER (9.4)
  )
  Whether the table c. and its sample data are like
  
  FIELDNAME DATACHR REVISIONNO DATANUM PRODUCTIONCODE
  AB No. Nch 1 100 0
  AB No. Nch 1 108 0
  AB No.-of-Nch 1 1 0
  Ass42 teased dishes 1 0
  Ass42 BTDrumNo1 BTD-051 1 0
  AB53 BTDrumNo1 BTD-051 104 0
  
  
  CREATE TABLE materialcode
  (
  UPPERMATERIALCODE CHAR (20),
  PROCESSNO TANK (3),
  PROCESSADOPTDATE TANK (9)
  
  )
  Whether the table b. and its sample data are like
  
  UPPERMATERIALCODE PROCESSNO PROCESSADOPTDATE
  Ass42 1 20120717
  AB53 108 20121121
  111 20111104 AS05
  104 20120928 AS30
  
  
  CREATE TABLE BUILDINGSCHEDULEMASTER
  (
  BUILDINGSIZE TANK (5 BYTES),
  GTCODE FLOAT (4 BYTES),
  TIREPERCART NUMBER (3.0).
  BUILDINGLOT NUMBER (3.0)
  )
  Whether table and its sample data are like
  
  BUILDINGSIZE GTCODE TIREPERCART BUILDINGLOT
  AB42A 4 12 ass42
  AB53A 4 88 AB53
  AS30A AS30 4 80
  BF03A 1 90 BF03
  
  
  Now, I need to get some data in these fields by using the suite of mappings.
  
  « PR_ » || Trim (a.BUILDINGSIZE) | ' / MPL2' Processid,--PRIMARY KEY
  TO_DATE (substr (NVL(b.PROCESSADOPTDATE,'19800101'), 7, 2) |) » /'|| SUBSTR (NVL(b.PROCESSADOPTDATE,'19800101'), 5, 2) | » /'|| SUBSTR (NVL(b.PROCESSADOPTDATE,'19800101'), 1, 4). (' 01:00:00 ',' dd-mm-yyyy hh24:mi:ss') starteff, to_date (SUBSTR (NVL(b.PROCESSADOPTDATE,'19800101'), 7, 2) |) » /'|| SUBSTR (NVL(b.PROCESSADOPTDATE,'19800101'), 5, 2) | » /'|| SUBSTR (NVL(b.PROCESSADOPTDATE,'19800101'), 1, 4). (' 01:00:00 ',' dd-mm-yyyy hh24:mi:ss') startDate.
  b.PROCESSNO revisioncodeid,
  a.BUILDINGLOT tirepercart ,
  CASE WHEN c.FIELDNAME = 'BTDrumNo1' then c.DATACHR end drumtype1.
  BOX WHEN (trim (C.DATACHR) = '1' and trim (C.FIELDNAME) = "No. Nch") and THEN click 'YES '.
  ANOTHER 'NO '.
  END may-BUG,
  case
  When c.FIELDNAME = 'Wavy'
  AND c.DATACHR = 'Use' THEN on 'YES '.
  When c.FIELDNAME = 'Wavy'
  AND c.DATACHR <>'Use' THEN 'NO '.
  fine wavy
  
  
  Now for the tables A and B I have the simple join as condition
  
  a.GTCODE = b.UPPERMATERIALCODE
  
  But I have to use two different left join conditions to join the table B and C to calculate the RESP fields according to the condition.
  Now, I have to use
  
  (trim (b.UPPERMATERIALCODE)). » -'|| Trim (b.PROCESSNO) = Trim (c.PRODUCTIONCODE) | » -'|| Trim (c.REVISIONNO)
  
  condition to get the data for the WAVY and DRUMTYPE1 fields.
  
  But at the same time, I have to use
  
  substr (Trim (b.UPPERMATERIALCODE), 1, 2). » -'|| Trim (b.PROCESSNO) = Trim (c.PRODUCTIONCODE) | » -'|| Trim (c.REVISIONNO)
  
  conition to derive the CHAFER field.
  
  And for this reason I am come so many duplicates and do not correct results for JUNEBUG, WAVY, DRUMTYPE1, and STARTDATE field.
  
  I need some advice as how can I do so that I get the correct results without duplicates for the PRIMARY KEY.
  
  Thank you
  
  Mahesh

  It would have been easier to answer if you gave us your example query that gives incorrect results - which both show us what you are doing and give us a starting point.
  Insert instructions for data, rather than only the data would have helped too, making it more quick to test this point.
  You do not show what you have done with your two outer join criteria, but I think you need to list TIREMASTERPROCESS_TEMP in your query twice, with alias names. Here's what I did:

  select a.BUILDINGSIZE, a.GTCODE, b.PROCESSNO ,  c1.FIELDNAME, c1.DATACHR, C2.FIELDNAME, C2.DATACHR
  , CASE WHEN c1.FIELDNAME='BTDrumNo1' then c1.DATACHR end drumtype1
  , CASE WHEN (trim(c2.DATACHR) = '1' and trim(c2.FIELDNAME) = 'No-of-Nch') THEN 'YES' ELSE 'NO' END CHAFER
  , case when c1.FIELDNAME='Wavy' AND c1.DATACHR='Use' THEN 'YES'
         when c1.FIELDNAME='Wavy' then 'NO' else null end wavy
  from BUILDINGSCHEDULEMASTER        A
  join materialcode b on a.GTCODE=b.UPPERMATERIALCODE
  left outer join TIREMASTERPROCESS_TEMP c1
     ON trim(b.UPPERMATERIALCODE)||'-'||trim(b.PROCESSNO) = trim(c1.PRODUCTIONCODE)||'-'||trim(c1.REVISIONNO)
  left outer join TIREMASTERPROCESS_TEMP c2
     ON substr(trim(b.UPPERMATERIALCODE),1,2)||'-'||trim(b.PROCESSNO) = trim(c2.PRODUCTIONCODE)||'-'||trim(c2.REVISIONNO)
  

  C1 is an alias for the TIREMASTERPROCESS_TEMP line which is necessary for drumtype1 and wavy, c2 is an alias for the TIREMASTERPROCESS_TEMP line required for JuneBug. There was some minor errors in your calculations of field I tried to correct.

  The results are:

  BUILD GTCO PRO FIELDNAME    DATACHR    FIELDNAME    DATACHR    DRUMTYPE1  CHA WAV
  ----- ---- --- ------------ ---------- ------------ ---------- ---------- --- ---
  AB53A AB53 108                         No-of-Nch    1                     YES
  AB42A AB42 1   BTDrumNo1    BTD-051    No-of-Nch    1          BTD-051    YES
  AB42A AB42 1   Wavy         NotUse     No-of-Nch    1                     YES NO
  AS30A AS30 104                                                            NO
  

  Hope this helps,
  David

• Seen the lock of brain on Left Join

  Always on Oracle 11.2.0.1.0
  
  Examples of data
  

  CREATE TABLE theEmails (
  RID NUMBER(10) PRIMARY KEY,
  employee NUMBER(6),
  emailType VARCHAR2(4),
  emailAddress VARCHAR2(50),
  emailActive number(1)
  )
  ;
  INSERT INTO theEmails VALUES(778437,231509,'OTHR','[email protected]',0);
  INSERT INTO theEmails VALUES(847168,231509,'WRK1','[email protected]',0);
  INSERT INTO theEmails VALUES(784690,231509,'OTHR','[email protected]',1);
  INSERT INTO theEmails VALUES(1459884,239868,'OTHR','[email protected]',0);
  INSERT INTO theEmails VALUES(1485781,239868,'OTHR','[email protected]',1);
  INSERT INTO theEmails VALUES(1485782,239868,'WRK1','[email protected]',0);
  INSERT INTO theEmails VALUES(1501028,231509,'WRK1','[email protected]',1);
  INSERT INTO theEmails VALUES(1492288,239868,'WRK1','[email protected]',1);
  INSERT INTO theEmails VALUES(1456785,239865,'OTHR','[email protected]',0);
  INSERT INTO theEmails VALUES(1489166,239865,'WRK1','[email protected]',0);
  INSERT INTO theEmails VALUES(1482668,239865,'WRK1','[email protected]',0);
  INSERT INTO theEmails VALUES(1482667,239865,'OTHR','[email protected]',1);
  
  
  
  CREATE TABLE thePeople (
  employee NUMBER(6),
  recordNum NUMBER(2),
  lastName VARCHAR2(20),
  firstName VARCHAR2(20)
  )
  ;
  
  INSERT INTO thePeople VALUES(231509,0,'P','Michelle');
  INSERT INTO thePeople VALUES(239868,0,'K','Dan');
  INSERT INTO thePeople VALUES(239865,0,'T','Mary');

  So, here is my SQL that does not work well:
  

  SELECT p.employee, p.recordNum, p.lastName, p.firstName, e.emailaddress, e.emailactive, e.emailtype
  FROM thePeople p LEFT JOIN theemails e ON p.employee = e.employee
  WHERE (e.emailactive = 1)
        AND
        (e.emailtype = 'WRK1')
        AND
        (e.emailaddress LIKE '%school.edu')
  ;

  AND this:
  

  SELECT p.employee, p.recordNum, p.lastName, p.firstName, e.emailaddress, e.emailactive, e.emailtype
  FROM thePeople p LEFT JOIN theemails e ON p.employee = e.employee
  WHERE (e.emailactive = 1 OR e.emailactive IS NULL)
        AND
        (e.emailtype = 'WRK1' OR e.emailtype IS NULL)
        AND
        (e.emailaddress LIKE '%school.edu' OR e.emailaddress IS NULL)

  Both give me this:
  
  employee | record | lastName. firstName. emailAddress. emailActive | emailType
  231509 | 0 | P     | Michelle | [email protected] | 1. WRK1
  
  
  But what I need is the following:
  employee | record | lastName. firstName. emailAddress. emailActive | emailType
  239868 | 0 | K     | Dan |      |     |
  231509 | 0 | P     | Michelle | [email protected] | 1. WRK1
  239865 | 0 | T     | Mary |          |
  
  
  Email from Dan is not included, because his "wrk1" email does not stop it school.edu
  E-mail Mary is not included, because his 'wrk1' email is not active.
  However, I need to come back with their numbers of name registration and employee.
  
  
  Once it works, I'll add employee = number in the WHERE clause, because I only need to find a person both for the actual production.
  
  What I am doing wrong?
  
  Thank you!
  
  Michelle

  Hello

  WHERE is the filtering actually record the selection.

  In your case, you might want to include in your LEFT OUTER JOIN ON conditions:

  SELECT p.employee, p.recordnum, p.lastname, p.firstname
       , e.emailaddress, e.emailactive, e.emailtype
    FROM    thepeople p
         LEFT JOIN
            theemails e
         ON     p.employee = e.employee
            AND e.emailactive = 1
            AND e.emailtype = 'WRK1'
            AND e.emailaddress LIKE '%school.edu';
  

  If you put them in your JOIN conditions actually join Chronogram of table theemails that adapt to the conditions of membership.
  As you use LEFT [OUTER] JOIN unmatched chronogram thepeople table will be displayed even when the theemail table column appears with a NULL value in the output files

  I hope that I was able to unlock your brain. :-)

  Kind regards.
  Al

  Published by: Alberto Faenza on 25 October 2012 16:58
  Additional note added

• Left join with three-table join query

  I am trying to create a query that left me speechless. Most of the query is simple enough, but I have a problem I do not know how to solve.
  
  Background:
  We have stock stored in i_action.
  
  We have the attributes available for each type of action. The attributes available for each action are described in shared_action_attribute. Each type of action can have three attributes or none at all.
  
  We have the values stored for the attributes in i_attribute_value.
  
  An example says:
  We have a transfer action (action_code B4). The action of B4 entry into i_action records the fact that the transfer took place and the date at which he spoke. The attributes available for a transfer action are the function code receiver, the receiving unit number and the reason of transfer code. These types of attributes available and their order are stored in shared_action_attribute. The actual values of the attributes for a specific action of transfer are stored in i_attribute_value.
  
  Now i_action and i_attribute_value can be connected directly in action_seq in i_action and ia_action_seq in i_attribute_value. A left join on these two tables provides results for all actions (including actions that have no attributes) and assign values (see Query 1 below).
  
  There are two questions. First of all, I want only the first two attributes. To specify the attributes of the first two, I also i_attribute_value a link to shared_action_attribute (which is where the order is stored). I can build a simple query (without the left join) which connects the three tables, but then shares without attributes would be excluded from my result (see Query 2 below).
  
  The second problem is that I'd actually a row returned for each action with first_attribute and second_attribute in the form of columns instead of two lines.
  
  The final query will be used to create a materialized view.
  
  Here are the tables and examples of what is stored in the:
  
  TABLE i_action
  Name Type
  ----
  ACTION_SEQ NUMBER (10)
  DATE OF ACTION_DATE
  ACTION_CODE VARCHAR2 (3)
  VARCHAR2 (1) DELETED
  
  EXAMPLE OF LINES
  ACTION_SEQ ACTION_DATE DELETED ACTION_CODE
  ----
  45765668 9 OCTOBER 09 B2 HAS
  45765670 9 OCTOBER 09 BA HAS
  45765672 B6 9 OCTOBER 09A
  45765673 9 OCTOBER 09 B4 HAS
  45765674 9 OCTOBER 09 G1 HAS
  45765675 9 OCTOBER 09 M3 HAS
  
  TABLE i_attribute_value
  Name Type
  ---
  IA_ACTION_SEQ NUMBER (10)
  SACTATT_SACT_CODE VARCHAR2 (3)
  SACTATT_SAT_TYPE VARCHAR2 (3)
  VARCHAR2 VALUE (50)
  
  EXAMPLE OF LINES
  IA_ACTION_SEQ SACTATT_SACT_CODE SACTATT_SAT_TYPE VALUE
  ----
  45765668 B2 COA 37 B
  45765670 BA ROA D
  45765670 BA ROR P
  45765672 B6 CAT C
  B4 45765673 RFC E
  45765673 B4 TRC P
  B4 45765673 RUN 7
  45765674 G1 SS 23567
  G1 45765674 ASG W
  
  TABLE shared_action_attribute
  Name Type
  ---
  SACT_CODE VARCHAR2 (3)
  SAT_TYPE VARCHAR2 (3)
  ORDER NUMBER (2)
  TITLE VARCHAR2 (60)
  
  EXAMPLE OF LINES
  SACT_CODE SAT_TYPE UNDER THE ORDER
  ----
  B2 ACO 1 Office code
  BA ROR 1 reason to re-open
  Authority of BA ROA 2 reopen
  B6 CAT 1 category
  B4 RFC 1 reception function code
  B4 RUN 2 receives the unit code
  B4 TRC 3 transfer of reason code
  Sequence of G1 SS 1 personal
  Reason for G1 ASG 2 assignment
  
  QUERY 1:
  It's my current query as well as its results. Most are select simple but only one column is filled using the function analytic last_value (thank you guys). The last column in the view sub stores the value of the attribute. What I want is to replace this single column with two columns named first_attribute and second_attribute and eliminate all other attributes.
  SELECT ia.action_seq, ia.action_date, ia.action_code cod,
  NVL
  (LAST_VALUE (CASE
  WHEN ia.action_code = "G1".
  AND iav.sactatt_sat_type = 'SS '.
  THEN THE VALUE
  WHEN ia.action_code IN ('A0', 'A1')
  THEN '67089'
  END IGNORE NULLS
  ) OVER (PARTITION BY ia.ici_charge_inquiry_seq ORDER BY ia.action_date,
  IA.serial_number, ia.action_seq),
  '67089'
  ) staff_seq,.
  value
  From i_action LEFT JOIN i_attribute_value iav AI
  ON iav.ia_action_seq = ia.action_seq
  WHERE ia.deleted = 'A ';
  
  ACTION_SEQ ACTION_DA COD STAFF_SEQ VALUE
  ----
  45765668 9 OCTOBER 09 B2 67089 37 B
  45765670 9 OCTOBER 09 BA D 67089
  45765670 9 OCTOBER 09 BA 67089 P
  45765672 9 OCTOBER 09 B6 67089 C
  45765673 9 OCTOBER 09 B4 67089 E
  45765673 9 OCTOBER 09 B4 67089 P
  45765673 9 OCTOBER 09 67089 7 B4
  45765674 9 OCTOBER 09 23567 23567 G1
  45765674 9 OCTOBER 09 G1 23567 W
  45765675 9 OCTOBER 09 M3 23567
  
  QUERY 2:
  This query is limited to the first two attributes but he also filed actions which have no attributes, and it creates still several lines for each action instead of a single line with two columns for attributes.
  SELECT ia.action_seq, ia.action_date, ia.action_code cod,
  NVL
  (LAST_VALUE (CASE
  WHEN ia.action_code = "G1".
  AND iav.sactatt_sat_type = 'SS '.
  THEN THE VALUE
  WHEN ia.action_code IN ('A0', 'A1')
  THEN '67089'
  END IGNORE NULLS
  ) OVER (PARTITION BY ia.ici_charge_inquiry_seq ORDER BY ia.action_date,
  IA.serial_number, ia.action_seq),
  '67089'
  ) staff_seq,.
  value
  OF shared_action_attribute saa, ims_action AI, ims_attribute_value iav
  WHERE iav.ia_action_seq = ia.action_seq
  AND iav.sactatt_sact_code = saa.sact_code
  AND iav.sactatt_sat_type = saa.sat_type
  AND saa.display_order IN ('1 ', ' 2')
  AND ia.deleted = 'A ';
  
  ACTION_SEQ ACTION_DA VALUE OF COD
  ----
  45765668 9 OCTOBER 09 B2 67089 37 B
  45765670 9 OCTOBER 09 BA D 67089
  45765670 9 OCTOBER 09 BA 67089 P
  45765672 9 OCTOBER 09 B6 67089 C
  45765673 9 OCTOBER 09 B4 67089 E
  45765673 9 OCTOBER 09 67089 7 B4
  45765674 9 OCTOBER 09 23567 23567 G1
  45765674 9 OCTOBER 09 G1 23567 W
  
  I found it quite complex to try to write - I hope that I was clear.
  
  Thank you very much!

  Hello

  You can use an alias for column (such as staff_seq) in the ORDER BY. Unfortunately, it's the only place where you can use it in the same query, where it was defined.
  You can use it anywhere in the super-requetes, however, so you can still work around this problem in assigning the aliases in a subquery and GROUP BY (or other) in a Super query, like this:

  WITH   ungrouped_data      AS
  (
      SELECT ia.action_seq, ia.action_date, ia.action_code,
            NVL
                (LAST_VALUE (CASE
                                WHEN ia.action_code = 'G1'
                                 AND sactatt_sat_type = 'SS'
                                   THEN VALUE
                                WHEN ia.action_code IN ('A0', 'A1')
                                   THEN '67089'
                             END IGNORE NULLS
                            ) OVER (PARTITION BY ia.ici_charge_inquiry_seq ORDER BY ia.action_date, ia.action_seq),
                 '67089'
                )staff_seq,
             (CASE
                WHEN display_order = '1'
                THEN VALUE
             END) first_attribute,
             (CASE
                WHEN display_order = '2'
                THEN VALUE
             END) second_attribute
        FROM i_action ia
        LEFT JOIN i_attribute_value iav
             ON iav.ia_action_seq = ia.action_seq
        LEFT JOIN shared_action_attribute
             ON sactatt_sact_code = sact_code
           AND sactatt_sat_type = sat_type
       WHERE ia.deleted = 'A'
  )
  SELECT       action_seq
  ,       action_date
  ,       action_code
  ,       staff_seq
  ,       MIN (first_attribute)          AS first_attribute
  ,       MIN (second_attribute)     AS second_attribute
  FROM       ungrouped_data
  GROUP BY  action_seq
  ,       action_date
  ,       action_code
  ,       staff_seq
  ;
  

  There are other alternatives for special cases, but none of them work in this particular case.

• Left Join with a subquery

  Hey everybody!
  
  I have some difficulty in trying to create a SQL query.
  
  I have the following tables:
  
  -Table Products: Product list
  CREATE TABLE (PRODUCTS)
  PRODUCT_ID INTEGER NOT NULL,
  KEY ("PRODUCT_ID") PRIMARY CONSTRAINT 'PK_PRODUCTS '.
  );
  
  -Orders table: Decreed this product
  CREATE TABLE ORDERS)
  ORDER_ID INTEGER NOT NULL,
  PRODUCT_ID INTEGER NOT NULL, / * FK PRODUCTS (PRODUCT_ID) * /.
  KEY ("ID_PEDIDO") PRIMARY CONSTRAINT 'PK_ORDERS '.
  );
  
  -Orders_Approval table: register when an order is approved
  CREATE TABLE ORDERS_APPROVAL)
  ORDER_ID INTEGER NOT NULL, / * FK ORDERS (ORDER_ID) * /.
  APPROVAL_DATE DATE NOT NULL,
  USER_ID INTEGER,
  KEY("ORDER_ID","APPROVAL_DATE") PRIMARY CONSTRAINT 'PK_ORDERS_APPROVAL '.
  );
  
  -Table Orders_ApprovalData: Register more details on the amenities of the order
  CREATE TABLE ORDERS_APPROVAL_DETAILS)
  APPROVALDET_ID INTEGER NOT NULL,
  ORDER_ID INTEGER NOT NULL, / * FK ORDERS_APPROVAL (ORDER_ID) * /.
  APPROVAL_DATE DATE NOT NULL, / * FK ORDERS_APPROVAL (APPROVAL_DATE) * /.
  KEY ("APPROVALDET_ID") PRIMARY CONSTRAINT 'PK_ORDERS_APPROVAL_DETAILS '.
  );
  
  The thing is that an order may OU cannot be approved. Thus, there could be no record of an order in ORDERS_APPROVAL.
  
  In addition, an order can have more than one approval.
  
  Here is the data that I use:
  
  /===========\
  | PRODUCTS |
  |===========|
  | PRODUCT_ID |
  |===========|
  | 1.
  |===========|
  | 2.
  |===========|
  | 3.
  \===========/
  
  /========================\
  | ORDERS |
  |========================|
  | ORDER_ID | PRODUCT_ID |
  |===========|============|
  | 27. 1.
  |===========|============|
  | 28. 2.
  |===========|============|
  | 29. 3.
  \===========|============/
  
  /=========================================\
  | ORDERS_APPROVAL |
  |=========================================|
  | ORDER_ID | APPROVAL_DATE | USER_ID.
  |===========|================|============|
  | 27. 01/10/2009 | 56.
  |===========|================|============|
  | 27. 04/10/2009 | 96.
  |===========|================|============|
  | 29. 03/10/2009 | 77.
  \===========|================|============/
  
  /=============================================\
  | ORDERS_APPROVAL_DETAILS |
  |=============================================|
  | APPROVALDET_ID | ORDER_ID | APPROVAL_DATE |
  |================|===========|================|
  | 1. 27. 01/10/2009 |
  |================|===========|================|
  | 2. 27. 04/10/2009 |
  |================|===========|================|
  | 3. 29. 03/10/2009 |
  \================|===========|================/
  I need to create a query that returns the data in all tables, but for orders with more than one certification,
  the query must return only the last record. It is also necessary return orders that have not yet been approved.
  
  My first query looked like this:
  
  SELECT
  *
  Of
  ORD IN ORDERS,
  PRODUCT PROD,
  ORDERS_APPROVAL ORDAPPROV,
  ORDERS_APPROVAL_DETAILS ORDAPPDET
  WHERE
  PROD. PRODUCT_ID = DSB. PRODUCT_ID
  AND ORDAPPROV. ORDER_ID (+) IS DSB. ORDER_ID
  AND ORDAPPDET. ORDER_ID (+) IS ORDAPPROV. ORDER_ID
  AND ORDAPPDET. DATA_APPROVAL_DATE (+) IS ORDAPPROV. APPROVAL_DATE
  
  PS: I use the Oracle 8 syntax because it's the server I'll use. In addition, I can't change the design of database.
  
  This query returns the data in table all. Because I used LEFT JOIN "(+)" to join the table orders_approval with the orders table, I am able to see even the orders with no approval.
  
  But this way, I see also two approvals for order 27 (with the data above, this order was approved twice). I need to reach my paintings in the last approval for each order is but should be a left join so I can see the orders with no approval.
  
  I tried this way:
  
  SELECT
  *
  Of
  ORD IN ORDERS,
  PRODUCT PROD,
  ORDERS_APPROVAL ORDAPPROV,
  ORDERS_APPROVAL_DETAILS ORDAPPDET
  WHERE
  PROD. PRODUCT_ID = DSB. PRODUCT_ID
  AND ORDAPPROV. ORDER_ID (+) IS DSB. ORDER_ID
  --Also added the line below
  AND ORDAPPROV. APPROVAL_DATE = (SELECT MAX (APPROVAL_DATE) OF ORDERS_APPROVAL WHERE THE ORDER_ID = ORDAPPROV. ORDER_ID)
  AND ORDAPPDET. ORDER_ID (+) IS ORDAPPROV. ORDER_ID
  AND ORDAPPDET. DATA_APPROVAL_DATE (+) IS ORDAPPROV. APPROVAL_DATE
  
  But in this way, as I am specifying I want only records with approval equal to date at the latest for each order, I do not see the order without approval.
  
  I tried again, with a left join like this:
  
  SELECT
  *
  Of
  ORD IN ORDERS,
  PRODUCT PROD,
  ORDERS_APPROVAL ORDAPPROV,
  ORDERS_APPROVAL_DETAILS ORDAPPDET
  WHERE
  PROD. PRODUCT_ID = DSB. PRODUCT_ID
  AND ORDAPPROV. ORDER_ID (+) IS DSB. ORDER_ID
  --Also added the line below
  AND ORDAPPROV. APPROVAL_DATE (+) = (SELECT MAX (APPROVAL_DATE) OF ORDERS_APPROVAL WHERE THE ORDER_ID = ORDAPPROV. ORDER_ID)
  AND ORDAPPDET. ORDER_ID (+) IS ORDAPPROV. ORDER_ID
  AND ORDAPPDET. DATA_APPROVAL_DATE (+) IS ORDAPPROV. APPROVAL_DATE
  
  But I get a syntax error. Oracle said that it is not possible to do a join with a subquery.
  
  If anyone knows how I can bring all this data?
  
  Thanks in advance.
  
  Published by: user9936895 on 05/10/2009 09:42: tried to fix the predetermined data schema.

  Hello

  Welcome to the forum!

  A scalar, such as subquery

  (SELECT MAX(APPROVAL_DATE) FROM ORDERS_APPROVAL WHERE ORDER_ID=ORDAPPROV.ORDER_ID)
  

  is allowed in a WHERE clause, but not as a condition of outer join, as you discovered.

  A workaround is the query expression so that it is not part of the outer join condition, which invlove in May (as in this case), adding another subquery. In the following query, the table of orders_approval in your original OF Division is replaced with online view, which uses the exact same scalar subquery in the WHERE clause. The online display will only contain one line per order_id (assuming that there is no link with for last approval_date).

  SELECT
       *
  FROM
       orders               ord,
       products          prod,
       (     -- Begin in-line view ORDAPPROV of last approval_date only
       SELECT     *       -- or list the columns needed in main query
       FROM     orders_approval         oa
       WHERE     approval_date = (
                      SELECT  MAX (approval_date)
                      FROM     orders_approval
                      WHERE     order_id     = oa.order_id
                      )
       )                ordapprov,
       orders_approval_details ordappdet
  WHERE
       prod.product_id               = ord.product_id
  AND     ordapprov.order_id (+)          = ord.order_id
  AND     ordappdet.order_id (+)          = ordapprov.order_id
  AND     ordappdet.approval_date (+)     = ordapprov.approval_date
  ;
  

  Thank you for including CREATE TBAL statements; It is very useful.
  You want to be even more useful in the future? Post INSERT statements for the sample data. That will make it easier for people to respond and easier for them to test their ideas.

• Including only in rows on a left join

  Hi, consider the following (hypothetical example http://www.codinghorror.com/blog/archives/000976.html but a real problem more complex mirror I)
  
  create table test1 (identification number, name varchar2 (10), color varchar2 (10))
  create table test2 (identification number, name varchar2 (10), color varchar2 (10))
  
  insert into test1 values (1, 'pirate', 'black')
  insert into test1 values (2, 'monkey', 'black')
  insert into test1 values (3, 'ninja', 'black')
  insert into test1 values (4, 'spaghetti', 'black')
  
  
  insert into test2 values (1, 'Swede', 'black')
  insert into test2 values (2, 'pirate', 'red')
  insert into test2 values (3, 'darth', 'red')
  insert into test2 values (4, 'ninja', 'black')
  
  
  *****
  Running the following query left join as usual gives the expected result
  
  Select * from test1
  outer join test2 left test1.name = test2.name
  
  
  ID NAME ID NAME COLOR COLOR
  
  1 appropriate appropriate 2 black red
  3 black ninja 4 black ninja
  null null null 4 black spaghetti
  null null null 2 black monkey
  
  ****
  If I add a restriction on test2.colour, I lose the three rows...
  
  ID NAME ID NAME COLOR COLOR
  
  1 appropriate appropriate 2 black red
  
  1 selected lines
  
  ****
  I can get 2 of them by re - including NULL values...
  
  ID NAME ID NAME COLOR COLOR
  
  1 appropriate appropriate 2 black red
  null null null 4 black spaghetti
  null null null 2 black monkey
  
  3 selected lines
  
  But what I want is for the other row where test2.colour is black to come back that looks like this...
  
  ID NAME ID NAME COLOR COLOR
  
  1 appropriate appropriate 2 black red
  null null null 3 black ninja
  null null null 4 black spaghetti
  null null null 2 black monkey
  
  4 selected lines
  
  As you can see, rank test2.id = 4 is here with the id, name, and the null color. How can I do this?
  
  Thank you!
  :)
  
  Published by: user3809287 on 15-Sep-2008 03:56

  ****
  If I add a restriction on test2.colour, I lose the three rows...

  ID NAME ID NAME COLOR COLOR

  1 appropriate appropriate 2 black red

  1 selected lines

  No, you don't have to if you add the restriction to the join instead of a filter condition condition:

  Oracle9i Enterprise Edition Release 9.2.0.8.0 - Production
  With the Partitioning, OLAP and Oracle Data Mining options
  JServer Release 9.2.0.8.0 - Production
  
  Session altered.
  
  SQL>
  SQL> drop table test1;
  
  Table dropped.
  
  SQL> drop table test2;
  
  Table dropped.
  
  SQL>
  SQL> create table test1 (id number, name varchar2(10), colour varchar2(10));
  
  Table created.
  
  SQL> create table test2 (id number, name varchar2(10), colour varchar2(10));
  
  Table created.
  
  SQL>
  SQL> insert into test1 values (1, 'pirate','black');
  
  1 row created.
  
  SQL> insert into test1 values (2, 'monkey','black');
  
  1 row created.
  
  SQL> insert into test1 values (3, 'ninja','black');
  
  1 row created.
  
  SQL> insert into test1 values (4, 'spaghetti','black');
  
  1 row created.
  
  SQL>
  SQL> insert into test2 values (1, 'rutabaga','black');
  
  1 row created.
  
  SQL> insert into test2 values (2, 'pirate','red');
  
  1 row created.
  
  SQL> insert into test2 values (3, 'darth','red');
  
  1 row created.
  
  SQL> insert into test2 values (4, 'ninja','black');
  
  1 row created.
  
  SQL>
  SQL> commit;
  
  Commit complete.
  
  SQL>
  SQL> select * from test1
    2  left outer join test2 on test1.name = test2.name
    3  where test2.colour = 'red';
  
          ID NAME       COLOUR             ID NAME       COLOUR
  ---------- ---------- ---------- ---------- ---------- ----------
           1 pirate     black               2 pirate     red
  
  SQL>
  SQL> select * from test1
    2  left outer join test2 on test1.name = test2.name and test2.colour = 'red';
  
          ID NAME       COLOUR             ID NAME       COLOUR
  ---------- ---------- ---------- ---------- ---------- ----------
           1 pirate     black               2 pirate     red
           4 spaghetti  black
           3 ninja      black
           2 monkey     black
  
  SQL>
  

  You can read complete Gints Plivna notes on the joints here:

  http://www.gplivna.EU/papers/sql_join_types.htm

  Kind regards
  Randolf

  Oracle related blog stuff:
  http://Oracle-Randolf.blogspot.com/

  SQLTools ++ for Oracle (Open source Oracle GUI for Windows):
  http://www.sqltools-plusplus.org:7676 /.
  http://sourceforge.NET/projects/SQLT-pp/

• LEFT JOIN question

  Hello

  Here's 2 queries. First one help for the joints and the second query uses a join ancient. The first works very well and I tried to write the 2nd one to work exactly like the first. I don't know where to put the sub condition sysdate in 2nd query such that there operator.

    AND trunc(SYSDATE) BETWEEN hapf.effective_start_date(+) AND
         hapf.effective_end_date(+)
  

  1.

  SELECT papf.full_name,
         hapf.name position_name
    FROM per_people_x       papf,
         per_assignments_x  pax,
         hr_all_positions_f hapf
   WHERE papf.person_id = pax.person_id
         AND pax.position_id = hapf.position_id(+)
         AND trunc(SYSDATE) BETWEEN hapf.effective_start_date(+) AND
         hapf.effective_end_date(+)
  

  2.

  SELECT papf.full_name,
         hapf.name position_name
    FROM per_people_x      papf,
         per_assignments_x pax
    LEFT JOIN hr_all_positions_f hapf
      ON (pax.position_id = hapf.position_id)
   WHERE papf.person_id = pax.person_id
  

  Any suggestion is appreciated.

  Thank you

  KK

  SELECT papf.full_name,

  hapf. Name position_name

  THE women's wear per_people_x

  JOIN THE

  Pax per_assignments_x

  ON)

  PAPF.person_id = pax.person_id

  )

  LEFT JOIN

  hr_all_positions_f hapf

  ON)

  Pax.POSITION_ID = hapf.position_id

  AND

  trunc (sysdate) BETWEEN hapf.effective_start_date AND hapf.effective_end_date

  )

  /

  SY.

• LEFT JOIN DUPLICATION PROBLEM

  Hello

  I'm having a problem with the left join query, when I join table a two table based on column task1 I get duplicate in table1.task1, table1.price.

  Table1. Task1	Table1. Price	table2. Task1	table2. Resource	table2. Price
  001	100	001	A	50
  001	100	001	B	250

  How can I make a request to get a result as below.

  Table1. Task1	Table1. Price	table2.Task2	table2. Resource	table2. Price
  001	100	001	A	50
  		001	B	250

  Thank you.

  Note that your query uses an inner join. Your original question mentioned a join left, generally interpreted as meaning a left OUTER join.

  Anyway, according to Frank, you can use the BREAK command in SQL * Plus for the goal sought through formatting. You can also use an analytical function as Roger suggests. I think ROW_NUMBER() might do the trick, but we must be clear about the criteria for partitioning and ordering the results, for example

  WITH table1 AS (
    SELECT '001' AS task1
         , 100 AS price
    FROM   dual
  ), table2 AS (
    SELECT '001' AS task1
         , 'A' AS resources
         , 50 AS price
    FROM   dual
    UNION ALL
    SELECT '001' AS task1
         , 'B' AS resources
         , 250 AS price
    FROM   dual
  )
  SELECT DECODE(ROW_NUMBER() OVER (PARTITION BY t1.task1, t1.price ORDER BY t2.resources, t2.price),1,t1.task1) AS task1_alt
       , DECODE(ROW_NUMBER() OVER (PARTITION BY t1.task1, t1.price ORDER BY t2.resources, t2.price),1,t1.price) AS price_alt
       , t2.task1 AS task_with_resource
       , t2.resources
       , t2.price
  FROM   table1 t1
  INNER JOIN table2 t2
  ON     t1.task1 = t2.task1
  ORDER BY t1.task1, t1.price, t2.resources, t2.price;