Select from function return rowtype @ db link

Hi, I am trying to select a row from a function which return a row type @ another DB.
First DB
==========
Function: fn_something(params) return mytable@snddb%rowtype
DB Link: snddb
Second DB
==========
Table: mytable (Column: mycol)
I have tried to execute query
select fn_something(params) from dual
OR
select mycol from table(fn_something(params))
but I have got "invalid datatype" error.
What would be the correct syntax for my query? Thanks.

Hi Andy,
Thanks for your reply.
My function itself has no problem and has been using for a while.
The function return a rowtype(a table row) and it is using in a stored procedure.
It could be done in stored procedure but I would like to know if I could call it directly from a query.
I have just read another thread
Help - Oracle function w/RETURN VIEW_NAME%ROWTYPE
which is quite similar to mine
but I am trying to get a column value from the function and execute an insert statement.
Could I do it without wrapped by begin end ?

Similar Messages

Select * from FUNCTION(a,b,c,...)

Hi,
Our client asked us not to use VIEW in our entire application!
After wasting one week negotiating with them, I am looking for some alternatives for views to simplify our complex queries.
Can I use functions instead of views in select statements? Could you give me a sample for SCOTT.EMP table? I am looking for something like:
SELECT * FROM UDF_FUNCTION(123)
Are functions a good alternative for views? What are my other options?
Regards,
Alan

You can select from a function if you create a "pipelined function". I would definitely use a view if I could, since a pipelined function will almost certainly not perform as well as a well-tuned view. Pipelined functions can be useful for ETL transformations, but are probably not the best choice for data access routines to deliver results to an end-user.
Anyway, here is a simple example, using the SCOTT sample schema:
SQL> create type EmpType as object
2 (
3 empno number(4),
4 ename varchar2(10),
5 sal number(7,2)
6 );
7 /
Type created
SQL> create type EmpTabType as table of EmpType;
2 /
Type created
SQL> create or replace function EmpFunc(p_in_sal IN NUMBER) return EmpTabType PIPELINED IS
2    l_emp_rec EmpType := EmpType(NULL,NULL,NULL);
3 begin
4    for emp_rec in (select * from emp where sal > p_in_sal) loop
5      l_emp_rec.empno := emp_rec.empno;
6      l_emp_rec.ename := emp_rec.ename;
7      l_emp_rec.sal := emp_rec.sal;
8      PIPE ROW( l_emp_rec );
9    end loop;
10    return;
11 end EmpFunc;
12 /
Function created
SQL> select * from TABLE( EMPFUNC( 2000 ) );
     EMPNO ENAME             SAL
      7566 JONES            2975
      7698 BLAKE            2850
      7782 CLARK            2450
      7788 SCOTT            3000
      7839 KING             5000
      7902 FORD             3000
6 rows selected
SQL>

Select from Function PIPELINED on PL/SQL Block

Hi folks,
I have a PL/SQL Block and I need this block to show me in a grid the rows from a variable table (Type table) that I have declared.
I don´t want to create Database objects, that's why I haven´t created tableTest like a type object in the database.
Example:
Declare
type tableTest is table of varchar2(500);
function test(indName IN VARCHAR2) return tableTest PIPELINED as
begin
pipe row (indName);
pipe row ('teste2');
return
end;
begin
-- HERE I NEED SHOW ROWS RETURNED BY FUNCTION WAS DECLARED;
-- Select * from TABLE(test('NAME'));
end;

You declare a pipelined table function by specifying the PIPELINED keyword. Pipelined functions can be defined at the schema level with CREATE FUNCTION or in a package. The PIPELINED keyword indicates that the function returns rows iteratively. The return type of the pipelined table function must be a supported collection type, such as a nested table or a varray. This collection type can be declared at the schema level or inside a package. Inside the function, you return individual elements of the collection type. The elements of the collection type must be supported SQL datatypes, such as NUMBER and VARCHAR2. PL/SQL datatypes, such as PLS_INTEGER and BOOLEAN, are not supported as collection elements in a pipelined function.
http://download-uk.oracle.com/docs/cd/B19306_01/appdev.102/b14261/tuning.htm

Xmlgen.getxml("select * from table") returns null pointer exception

I am running oracle 8i on solaris server and clinet on windows
NT and i am this select statement
select xmlgen.getxml("select * from table") from dual ,its
returning null pointer exception,i have tried it through
jdbc,even then its returning xml as
<?xml version = '1.0'?>
<ERROR>java.lang.NullPointerException</ERROR>
can any body tell me the error.Help will be really appreciated.I
need an urgent response,if some one can guide me please.
My email is [email protected],if you can give me a quick
response on this email,your effot will be appreciated.
thanks
Masood

What is actually throwing the NullPointerException? rs.getMetaData() or table.setModel()?

'select count(*) from x' returns 5460 rows and 'Select * from x' returns 0 rows

As you can see in the next lines something is wrong in my Oracle (8.0.6 on Win NT 4.0 sp 6a) :
Oracle8 Enterprise Edition Release 8.0.6.0.0 - Production
With the Partitioning option
PL/SQL Release 8.0.6.0.0 - Production
SQLWKS> SELECT * FROM V_TERRA_TE;
PERIOD_DATE PERIOD_TIME TERRARCV TERRASND TERCV TESND
0 rows selected.
SQLWKS> SELECT COUNT(*) FROM V_TERRA_TE;
COUNT(*)
5460
1 row selected.
V_TERRA_TE is a complex join of 5 Tables :
CREATE OR REPLACE VIEW "ACTUATE".V_TERRA_TE AS Select to_date(to_char(p1.period_date,'YYYYMMDD'),'YYYYMMDD') "PERIOD_DATE",
to_date(to_char(p1.period_time,'HH24:MI'),'HH24:MI') "PERIOD_TIME",
to_number(p1.caudalrcv + ((p3.caudalrcv + p4.caudalrcv)*(confterrate.conexadslterra/confterrate.conexadsltotal))) "TERRARCV",
to_number(p1.caudalsnd + ((p3.caudalsnd + p4.caudalsnd)*(confterrate.conexadslterra/confterrate.conexadsltotal))) "TERRASND",
to_number((p2.caudalrcv * confterrate.pctinfonegocio) + ((p3.caudalrcv + p4.caudalrcv)*(confterrate.conexadslte/confterrate.conexadsltotal)))"TERCV",
to_number((p2.caudalsnd * confterrate.pctinfonegocio) + ((p3.caudalsnd + p4.caudalsnd)*(confterrate.conexadslte/confterrate.conexadsltotal)))"TESND"
from p1,p2,p3,p4,confterrate
where (p1.period_date=p2.period_date)and
(p1.period_date=p3.period_date)and
(p1.period_date=p4.period_date)and
(p1.period_time=p3.period_time)and
(p1.period_time=p4.period_time)and
(p1.period_time=p2.period_time)and
to_char(p1.period_date,'MMYYYY')=to_char(confterrate.period_datetime,'MMYYYY');
I think that some not reported error happens in the select * with some temporary space or similar but only the message '0 rows selected' is displayed (instead the real error)
Could somebody help me ?
Thanks in advance
Francisco

Forcing the Join/sort to be made on Disk (not on memory) the problem not happens. This demostrate that ORACLE has a VERY IMPORTANT BUG : It returns 0 rows wich is false.
To force it to work on disk i use this parameters :
alter session set sort_area_size=0
alter session set hash_join_enabled=false
Note : probably is not the best combination or use of parameters, but using it the query works as espected.

Creating an LOV from function returning Varray

I have a select statement which runs in PLSQL and runs in the SQL Commands of SQL Workshop, but when I try to use the select statement as an LOV, I keep getting "LOV query invalid, a display and a return value are needed, the columns names need to be different. if your query contains an in-line query, the first FROM clause.....must not belong to the in-line query." My select statement is as follows:
select a.obj, a.ename from THE (select cast(getxmlnodes('filename.xml') as NodeTableType2 from dual) a;
The function getxmlnodes reads an xml file and returns the nodes from my xml file. It works fine from SQLPlus etc. but the LOV balks. Can anyone out there help? Thanks

For anyone who sees this. My select statement didn't work because I had put a
';' at the end of the statement. I accidentally left off the ; in one of my tests and the select statement worked like a charm. Everyone likes to have a person answer their own question.

Select from function with named parameters doesn't work

Hello,
I'm trying to execute the next sql statement:
SELECT mypack.getvalue(user_id => 231, status => 'closed') AS someAlias FROM DUAL;
I'm getting the next Error:
Error: ORA-00907: missing right parenthesis
But the next works fine:
SELECT mypack.getvalue(231,'closed') AS someAlias FROM DUAL;
What I'm doing wrong?
Is there a way to call a Function and return it's result as a single-row query?
Thanks for any suggestions.

Thanks for your answers.
Just want to explain what I want to accomplish:
I want to create PL/SQL statement which:
1) Calls Function in named notation way;
2) Returns a query which contains a single row - a Function result.
I know in Transact-SQL I can accomplish this the next way:
DECLARE @return_value INT
EXEC[myproc]
@id=2,
@status='ok',
@ret_param=@return_value OUTPUT
SELECT @return_value AS my_return_value
The last SELECT call returns a query with one row in it.
How can I do the same in Oracle?
Thanks a lot!

Selecting from a function that returns a sys refcursor or an alternative

I have a query that returns a resultset of three columns, namely SSN,PAID_YEAR and PAID_TOTAL. From this query I can:
Create a view and then query it.
Create a function and return resultset
If I go the first way a simple query like the following takes more than 20 seconds:
SELECT PAID_YEAR,PAID_TOTAL FROM VIEW_1 WHERE SSN=12345678912882;
I know that is because when I query a view the engine first brings all the rows of the view and then it filters the rows for the criteria supplied.
If I go the second way I can send a parameter and make the engine look only for those rows that match the condition and return the recordset. But I do not know how to then SELECT from that returned resultset. I took a look at pipelined tables but didn't quite get how to benefit them. So my ultimate question is if it's somehow possible to select from the resultset that is returned from a function like this:
SELECT * FROM FUNCTION_1(12132323232).
If yes, then how, if no, what would be an alternative way?

I know that is because when I query a view the engine first brings all the rows of the view and then it filters the rows for the criteria supplied.
No - you don't 'know that' because it isn't true. Just check the explain plan for yourself. Oracle can still use any appropriate indexes so that only the rows needed are returned.
So my ultimate question is if it's somehow possible to select from the resultset that is returned from a function like this:
SELECT * FROM FUNCTION_1(12132323232).
No - you can't do it like that. You have to use the TABLE function to treat the function result set as a table:
'SELECT * FROM TABLE(FUNCTION_1(12132323232)).
-- type to match emp record
create or replace type emp_scalar_type as object
(EMPNO NUMBER(4) ,
   ENAME VARCHAR2(10),
   JOB VARCHAR2(9),
   MGR NUMBER(4),
   HIREDATE DATE,
   SAL NUMBER(7, 2),
   COMM NUMBER(7, 2),
   DEPTNO NUMBER(2)
-- table of emp records
create or replace type emp_table_type as table of emp_scalar_type
-- pipelined function
create or replace function get_emp( p_deptno in number )
return emp_table_type
PIPELINED
as
   TYPE EmpCurTyp IS REF CURSOR RETURN emp%ROWTYPE;
    emp_cv EmpCurTyp;
    l_rec emp%rowtype;
begin
    open emp_cv for select * from emp where deptno = p_deptno;
    loop
      fetch emp_cv into l_rec;
      exit when (emp_cv%notfound);
      pipe row( emp_scalar_type( l_rec.empno, LOWER(l_rec.ename),
          l_rec.job, l_rec.mgr, l_rec.hiredate, l_rec.sal, l_rec.comm, l_rec.deptno ) );
    end loop;
    return;
end;
select * from table(get_emp(20))

Why does getdate() function return same time value for multiple select statements executed sequentially

When I run the following code
set nocount on
declare @i table(id int identity(1,1) primary key, sDate datetime)
while((select count(*) from @i)<10000)
begin
insert into @i(sDate) select getdate()
end
select top 5 sDate, count(id) selectCalls
from @i
group by sDate
order by count(id) desc
I get the following results.
sDate selectCalls
2014-07-30 14:50:27.510 406
2014-07-30 14:50:27.527 274
2014-07-30 14:50:27.540 219
2014-07-30 14:50:27.557 195
2014-07-30 14:50:27.573 170
As you can see the select getdate() function returned same time up to the milisecon 406 time for the first date value. This started happening when we moved our applications to a faster server with four processors. Is this correct or am I
going crazy?
Please let me know
Bilal

Observe that adding 2 ms is accurate only with datetime2. As noted above, datetime does not have ms resolution:
set nocount on
declare @d datetime, @i int, @d2 datetime2
select @d = getdate(), @i = 0, @d2 = sysdatetime()
while(@i<10)
begin
select @d2, @d, current_timestamp, getdate(), sysdatetime()
select @d = dateadd(ms,2,@d), @i = @i+1, @d2=dateadd(ms,2,@d2)
end
2014-08-09 08:36:11.1700395 2014-08-09 08:36:11.170 2014-08-09 08:36:11.170 2014-08-09 08:36:11.170 2014-08-09 08:36:11.1700395
2014-08-09 08:36:11.1720395 2014-08-09 08:36:11.173 2014-08-09 08:36:11.170 2014-08-09 08:36:11.170 2014-08-09 08:36:11.1700395
2014-08-09 08:36:11.1740395 2014-08-09 08:36:11.177 2014-08-09 08:36:11.170 2014-08-09 08:36:11.170 2014-08-09 08:36:11.1700395
2014-08-09 08:36:11.1760395 2014-08-09 08:36:11.180 2014-08-09 08:36:11.170 2014-08-09 08:36:11.170 2014-08-09 08:36:11.1700395
2014-08-09 08:36:11.1780395 2014-08-09 08:36:11.183 2014-08-09 08:36:11.170 2014-08-09 08:36:11.170 2014-08-09 08:36:11.1700395
2014-08-09 08:36:11.1800395 2014-08-09 08:36:11.187 2014-08-09 08:36:11.170 2014-08-09 08:36:11.170 2014-08-09 08:36:11.1700395
DATE/TIME functions:
http://www.sqlusa.com/bestpractices/datetimeconversion/
Kalman Toth Database & OLAP Architect
SQL Server 2014 Design & Programming
New Book / Kindle: Exam 70-461 Bootcamp: Querying Microsoft SQL Server 2012

Functions returning rowset

How can do a function which returns a Rowset?

SQL> EDIT return_rowset_pkg
CREATE OR REPLACE PACKAGE return_rowset_pkg
AS
TYPE rowset IS REF CURSOR;
FUNCTION return_rowset
RETURN rowset;
END return_rowset_pkg;
CREATE OR REPLACE PACKAGE BODY return_rowset_pkg
AS
FUNCTION return_rowset
RETURN rowset
IS
row_cursor rowset;
BEGIN
OPEN row_cursor
FOR
SELECT *
FROM your_table;
RETURN row_cursor;
END return_rowset;
END return_rowset_pkg;
SQL> START return_rowset_pkg
Package created.
Package body created.
SQL> VARIABLE g_ref REFCURSOR
SQL> EXEC :g_ref := return_rowset_pkg.return_rowset()
PL/SQL procedure successfully completed.
SQL> PRINT g_ref
null

ORA-01031 insufficient privileges error when selecting from a view

OK I think this might be a dumb question but I can't figure it out:
User John has been granted SELECT privilege (directly, not through a database role) to schema FRED.table1;
User John can issue select * from FRED.table1; and it works just fine.
User John has then been granted SELECT privilege (directly, not via a database role) to schema
MARK.view1;
MARK.view1 only selects from FRED.table1. No other tables are in the view1.
Schema MARK can successfully query the views. SELECT * FROM VIEW1 returns results.
I also checked the MARK schema to ensure that it has been granted SELECT on FRED.table1 directly which it has.
Now, when logged into schema John, I try SELECT * FROM MARK.VIEW1; and I get ORA-01031 insufficient privileges error.
I'm not sure how to troubleshoot this. If John is granted SELECT ANY TABLE, it of course works but I don't want John to have that powerful priv.
To recap, John has SELECT on both MARK.VIEW1 and the table which VIEW1 selects from (FRED.TABLE1).
John can select from FRED.TABLE1 no problem but receives a privilege error even though John has SELECT on MARK.VIEW1.
Any thoughts?
Oh, Oracle EE 10.2.0.4

JSebastian wrote:
OK I think this might be a dumb question but I can't figure it out:
User John has been granted SELECT privilege (directly, not through a database role) to schema FRED.table1;
User John can issue select * from FRED.table1; and it works just fine.
User John has then been granted SELECT privilege (directly, not via a database role) to schema
MARK.view1;
MARK.view1 only selects from FRED.table1. No other tables are in the view1.
Schema MARK can successfully query the views. SELECT * FROM VIEW1 returns results.
I also checked the MARK schema to ensure that it has been granted SELECT on FRED.table1 directly which it has.
Now, when logged into schema John, I try SELECT * FROM MARK.VIEW1; and I get ORA-01031 insufficient privileges error.
I'm not sure how to troubleshoot this. If John is granted SELECT ANY TABLE, it of course works but I don't want John to have that powerful priv.
To recap, John has SELECT on both MARK.VIEW1 and the table which VIEW1 selects from (FRED.TABLE1).
John can select from FRED.TABLE1 no problem but receives a privilege error even though John has SELECT on MARK.VIEW1.
Any thoughts?
Oh, Oracle EE 10.2.0.4Are you certain John has been granted select on Mark.view1 ? In order for that to work Mark would have to have been given select on Fred.table1 WITH GRANT OPTION ... otherwise the grant would fail and then John would not be able to select from the view because the grant was never successfully issued.
Here's a basic test case (which i think conforms to what you've outlined) to get it working.
drop user u1 cascade;
drop user u2 cascade;
drop user u3 cascade;
create user u1 identified by u1;
grant connect, resource to u1;
create user u2 identified by u2;
grant connect, resource, create view to u2;
create user u3 identified by u3;
grant connect, resource to u3;
connect u1/u1@orcl
create table test1 (col1 number);
grant select on test1 to u2 with grant option; --> this is the important part
grant select on test1 to u3;
connect u2/u2@orcl
create view test2 as select * from u1.test1;
grant select on test2 to u3;
connect u3/u3@orcl
select * from u2.test2;

Query from User_Scheduler_Jobs returns no rows in version 2.1.0.63

Using version 2.1.0.63, "select * from user_scheduler_jobs;" returns no rows. It works ok in version 1.5.5.
Edited by: User65423 on Feb 8, 2010 9:24 AM

Refer to this thread, where the cause seems to be due to SQL Developer not handling null interval data types.
bug? F9 gives no output dba_scheduler_jobs, version: 2.1.0.63.73
As mentioned in that thread you can use to_char on the interval columns as a work-around...
SELECT source,
destination,
comments,
flags,
job_name,
job_creator,
client_id,
global_uid,
program_owner,
program_name,
job_type,
job_action,
number_of_arguments,
schedule_owner,
schedule_name,
start_date,
repeat_interval,
end_date,
job_class,
enabled,
auto_drop,
restartable,
state,
job_priority,
run_count,
max_runs,
failure_count,
max_failures,
retry_count,
last_start_date,
TO_CHAR(last_run_duration),
next_run_date,
TO_CHAR(schedule_limit),
TO_CHAR(max_run_duration),
logging_level,
stop_on_window_close,
instance_stickiness,
system,
job_weight,
nls_env
FROM user_scheduler_jobs a;

Output of value returned from function in SELECT statement ??

Hi
I have created the below function
create or replace
FUNCTION jc_test
RETURN VARCHAR2
IS myrec VARCHAR2(270);
BEGIN
SELECT RPAD('*',270,'*')
INTO myrec
FROM DUAL ;
RETURN myrec;
END ;
and I executed the SELECT statement in Oracle SQL developer as a script
select
LENGTH(jc_test()) len
*,jc_test() rec*
from dual ;
I get exact output as below
LEN REC
270 ******************************************************************************************************************************************************************************************************************************************************************************
So here LEN is correctly shown as 270 characters but when I see the astrisk's (also appended with spaces ) its total length is 4000 characters
Can anyone give their thoughts on this.
Its fine if I run as normal, i mean not as a script.
Regards
jc
Edited by: JC on Jun 16, 2011 11:25 AM

Hi,
So here LEN is correctly shown as 270 charactersYes, correct
but when I see the astrisk's (also appended with spaces ) its total length is 4000 charactersNo buts. It is not padded. That is just how it is displayed by your tool. In SQL*Plus this is controlled by LINESIZE and COLUMN
SQL> create or replace function dummy return varchar2 as begin return null; end;
2 /
Function created.
SQL> select dummy from dual;
D
X
SQL> select dummy() from dual;
DUMMY()
SQL> col "dummy()" for a10
SQL> select dummy() from dual;
DUMMY()
SQL>P.S: It is a really bad idea to create your own wrapper functions, built-in functions. Hopefully you are not really doing that?
Regards
Peter

SOLVED: How can I use or call a function that returns %ROWTYPE?

Hi
edit: you can probably skip all this guff and go straight to the bottom...In the end this is probably just a question of how to use a function that returns a %rowtype. Thanks.
Currently reading Feuerstein's tome, 5th ed. I've downloaded and run the file genaa.sp, which is a code generator. Specifically, you feed it a table name and it generates code (package header and package body) that will create a cache of the specified table's contents.
So, I ran:
HR@XE> @"C:\Documents and Settings\Jason\My Documents\Work\SQL\OPP5.WEB.CODE\OPP5.WEB.CODE\genaa.sp"
749 /
Procedure created.
HR@XE> exec genaa('EMPLOYEES');which generated a nice bunch of code, viz:
create or replace package EMPLOYEES_cache is
    function onerow ( EMPLOYEE_ID_in IN HR.EMPLOYEES.EMPLOYEE_ID%TYPE) return HR.EMPLOYEES%ROWTYPE;
    function onerow_by_EMP_EMAIL_UK (EMAIL_in IN HR.EMPLOYEES.EMAIL%TYPE) return HR.EMPLOYEES%ROWTYPE;
    procedure test;
end EMPLOYEES_cache;
create or replace package body EMPLOYEES_cache is
    TYPE EMPLOYEES_aat IS TABLE OF HR.EMPLOYEES%ROWTYPE INDEX BY PLS_INTEGER;
    EMP_EMP_ID_PK_aa EMPLOYEES_aat;
    TYPE EMP_EMAIL_UK_aat IS TABLE OF HR.EMPLOYEES.EMPLOYEE_ID%TYPE INDEX BY HR.EMPLOYEES.EMAIL%TYPE;
    EMP_EMAIL_UK_aa EMP_EMAIL_UK_aat;
    function onerow ( EMPLOYEE_ID_in IN HR.EMPLOYEES.EMPLOYEE_ID%TYPE)
        return HR.EMPLOYEES%ROWTYPE is
        begin
            return EMP_EMP_ID_PK_aa (EMPLOYEE_ID_in);
        end;
    function onerow_by_EMP_EMAIL_UK (EMAIL_in IN HR.EMPLOYEES.EMAIL%TYPE)
        return HR.EMPLOYEES%ROWTYPE is
        begin
            return EMP_EMP_ID_PK_aa (EMP_EMAIL_UK_aa (EMAIL_in));
        end;
    procedure load_arrays is
        begin
            FOR rec IN (SELECT * FROM HR.EMPLOYEES)
            LOOP
                EMP_EMP_ID_PK_aa(rec.EMPLOYEE_ID) := rec;
                EMP_EMAIL_UK_aa(rec.EMAIL) := rec.EMPLOYEE_ID;
            end loop;
        END load_arrays;
    procedure test is
        pky_rec HR.EMPLOYEES%ROWTYPE;
        EMP_EMAIL_UK_aa_rec HR.EMPLOYEES%ROWTYPE;
        begin
            for rec in (select * from HR.EMPLOYEES) loop
                pky_rec := onerow (rec.EMPLOYEE_ID);
                EMP_EMAIL_UK_aa_rec := onerow_by_EMP_EMAIL_UK (rec.EMAIL);
                if rec.EMPLOYEE_ID = EMP_EMAIL_UK_aa_rec.EMPLOYEE_ID then
                    dbms_output.put_line ('EMP_EMAIL_UK lookup OK');
                else
                    dbms_output.put_line ('EMP_EMAIL_UK lookup NOT OK');
                end if;
            end loop;
        end test;
    BEGIN
        load_arrays;
    end EMPLOYEES_cache;
/which I have run successfully:
HR@XE> @"C:\Documents and Settings\Jason\My Documents\Work\SQL\EMPLOYEES_CACHE.sql"
Package created.
Package body created.I am now trying to use the functionality within the package.
I have figured out that the section
    BEGIN
        load_arrays;
    end EMPLOYEES_cache;
/is the initialization section, and my understanding is that this is supposed to run when any of the package variables or functions are referenced. Is that correct?
With that in mind, I'm trying to call the onerow() function, but it's not working:
HR@XE> select onerow(100) from dual;
select onerow(100) from dual
ERROR at line 1:
ORA-00904: "ONEROW": invalid identifier
HR@XE> select employees_cache.onerow(100) from dual;
select employees_cache.onerow(100) from dual
ERROR at line 1:
ORA-06553: PLS-801: internal error [55018]
HR@XE> select table(employees_cache.onerow(100)) from dual;
select table(employees_cache.onerow(100)) from dual
ERROR at line 1:
ORA-00936: missing expressionHe provides the code genaa.sp, and a very brief description of what it does, but doesn't tell us how to run the generated code!
Now, I have just done some googling, and it seems that what I am trying to do isn't possible. Apparently %ROWTYPE is PL/SQL, and not understood by SQL, so you can't call onerow() from sql. Correct?
So I try wrapping the call in an exec:
HR@XE> exec select employees_cache.onerow(100) from dual;
BEGIN select employees_cache.onerow(100) from dual; END;
ERROR at line 1:
ORA-06550: line 1, column 30:
PLS-00382: expression is of wrong type
ORA-06550: line 1, column 7:
PLS-00428: an INTO clause is expected in this SELECT statement
HR@XE> exec select table(employees_cache.onerow(100)) from dual;
BEGIN select table(employees_cache.onerow(100)) from dual; END;
ERROR at line 1:
ORA-06550: line 1, column 14:
PL/SQL: ORA-00936: missing expression
ORA-06550: line 1, column 7:
PL/SQL: SQL Statement ignored
HR@XE> exec employees_cache.onerow(100)
BEGIN employees_cache.onerow(100); END;
ERROR at line 1:
ORA-06550: line 1, column 7:
PLS-00221: 'ONEROW' is not a procedure or is undefined
ORA-06550: line 1, column 7:
PL/SQL: Statement ignoredNo joy.
Of course, now that I'm looking at it again, it seems that the way to go is indicated by the first error:
PLS-00428: an INTO clause is expected in this SELECT statement
So am I supposed to create a type of EMPLOYEES%ROWTYPE in a PL/SQL procedure, and the idea of this code, is that the first call to onerow() runs the initialiation code, which populates the cache, and all subsequent calls to onerow() (whether by my session or any other) will use the cache?
I've had a stab at this, but still, no joy:
create or replace procedure testcache is
    emp employees%rowtype;
    begin
        select employees_cache.onerow(100) from dual into emp;
        dbms_output.put_line('Emp id: ' || emp.employee_id);
    end testcache;
show errors
HR@XE> @testcache.sql
Warning: Procedure created with compilation errors.
Errors for PROCEDURE TESTCACHE:
LINE/COL ERROR
4/9      PL/SQL: SQL Statement ignored
4/54     PL/SQL: ORA-00933: SQL command not properly ended
HR@XE>Have a feeling this should be really easy. Can anybody help?
Many thanks in advance.
Jason
Edited by: 942375 on 08-Feb-2013 11:45

>
Ha, figured it out
>
Hopefully you also figured out that the example is just that: a technical example of how to use certain Oracle functionality. Unfortunately it is also an example of what you should NOT do in an actual application.
That code isn't scaleable, uses expensive PGA memory, has no limit on the amount of memory that might be used and, contrary to your belief will result in EVERY SESSION HAVING ITS OWN CACHE of exactly the same data if the session even touches that package.
Mr. Feuerstein is an expert in SQL and PL/SQL and his books cover virtually all of the functionality available. He also does an excellent job of providing examples to illustrate how that functionality can be combined and used. But the bulk of those examples are intended solely to illustrate the 'technical' aspects of the technology. They do not necessarily reflect best practices and they often do not address performance or other issues that need to be considered when actually using those techniques in a particular application. The examples show WHAT can be done but not necessarily WHEN or even IF a given technique should be used.
It is up to the reader to learn the advantages and disadvantages of each technicalogical piece and determine when and how to use them.
>
Now, I have just done some googling, and it seems that what I am trying to do isn't possible. Apparently %ROWTYPE is PL/SQL, and not understood by SQL, so you can't call onerow() from sql. Correct?
>
That is correct. To be used by SQL you would need to create SQL types using the CREATE TYPE syntax. Currently that syntax does not support anything similar to %ROWTYPE.
>
So am I supposed to create a type of EMPLOYEES%ROWTYPE in a PL/SQL procedure, and the idea of this code, is that the first call to onerow() runs the initialiation code, which populates the cache, and all subsequent calls to onerow() (whether by my session or any other) will use the cache?
>
NO! That is a common misconception. Each session has its own set of package variables. Any session that touches that package will cause the entire EMPLOYEES table to be queried and stored in a new associative array specifically for that session.
That duplicates the cache for each session using the package. So while there might be some marginal benefit for a single session to cache data like that the benefit usually disappears if multiple sessions are involved.
The main use case that I am aware of where such caching has benefit is during ETL processing of staged data when the processing of each record is too complex to be done in SQL and the records need to be BULK loaded and the data manipulated in a loop. Then using an associative array as a lookup table to quickly get a small amount of data can be effective. And if the ETL procedure is being processed in parallel (meaning different sessions) then for a small lookup array the additional memory use is tolerable.
Mitigating against that is the fact that:
1. Such frequently used data that you might store in the array is likely to be cached by Oracle in the buffer cache anyway
2. Newer versions of Oracle now have more than one cache
3. The SQL query needed to get the data from the table will use a bind variable that eliminates repeated hard parsing.
4. The cursor and the buffer caches ARE SHARED by multiple sessions globally.
So the short story is that there would rarely be a use case where ARRAYs like that would be preferred over accessing the data from the table.

How to get result of Select from stored function.

I need to get result of select from a stored function.
In the end of my stored function I makes final select (four columns).
How it can be retrived from function?

Hi,
A function can only return one value, but it sounds like you want to return 4 values.
The one value that you return can be a record, with many columns, such as a ROWTYPE, or a TYPE that you define.
You can return an XMLTYPE that has whatever elements you want.
You can write a procedure that has several OUT parameters. (You can have OUT parameters in a function, but a lot of people find that confusing.)
In very special circumstance, you might consider returning a string that is a delimited list of values, such as '7639,SMITH,,17-DEC-1980'.
Someoneelse has a good point.
We could give a better answer if you ask a specific question, like:
"I have this table ...
I want a function such that, if I call it with these parameters ... I get ...
but if I call it like this ... then I get ..."

Select from function return rowtype @ db link

Similar Messages

Maybe you are looking for