How to obtain SQL that don't use bind variables

Hi,
I'm trying to identify the SQL that should benefit from using bind variables.
First, I tried to obtain the common signature from all those sql calls, by using:
select * from (
select  force_matching_signature, count(1) from v$sql where force_matching_signature<>0 group by force_matching_signature order by 2 desc
) where rownum < 50;Then, I copied those values to clipboard and executed: select sql_text from v$sql where force_matching_signature=<<<copied_signature_value>>>;Now I want to make it automatically, and get just 1 occurrence of each SQL that is similar to others by using a query.
I've tried this:
select sql_text from
select sql_text, force_matching_signature, row_number() over (partition by force_matching_signature order by sql_text desc)rn from v$sql where force_matching_signature <>0
)where rn <2 and rownum < 10 order by force_matching_signature descBut this is not returning results by attending to the count(1) of the first query I used. How can I modify this so I get the results in order of 'importance'?
Thanks

And I said. First use order by, then use rownum. I didn't mention row_number. Also there should be no need to add any more columns.
Did you try it? Why didn't it work?
untested example
select * from (
   select sql_text from (select sql_text, force_matching_signature, row_number() over (partition by force_matching_signature order by sql_text desc) rn from v$sql where force_matching_signature != 0)
   where rn = 1
   order by force_matching_signature desc /* add any ordering you like here */
where rownum < 10  /* then filter on the first 10 results */If you want to order it in such a way that the statement that is found most times comes first then say so. However I do not see how to group in your case.
maybe like this
tested example
select * from (
   select cnt, sql_text
   from (select sql_text, force_matching_signature, row_number() over (partition by force_matching_signature order by sql_text desc) rn , count(*) over (partition by force_matching_signature) cnt
         from v$sql
         where force_matching_signature != 0)
   where rn = 1
   order by cnt desc, force_matching_signature desc /* add any ordering you like here */
where rownum < 10  /* then filter on the first 10 results */
;Edited by: Sven W. on Oct 11, 2012 2:49 PM
Edited by: Sven W. on Oct 11, 2012 2:51 PM
Edited by: Sven W. on Oct 11, 2012 2:56 PM -- added count column to the output

Similar Messages

  • How to formulate SQL that need to use some XML data in a clob?

    Hi,
    We just created a new 8i table that has some regular fields as well as a clob field that contains XML data. How could I bring back the value of a spcific element in the XML field and compare it against a regular field? For example, I want to make sure the value in field A is not the same as the value of a specific element in field b. Your help is very much appreciated.
    select * from X
    where X.a <> X.b.elementZ

    Depending on how complex your XML data is, you can write a simple function that does a string search of the CLOB and returns the value of the tag you are looking at. Then you can compare the value of the tag and the value of the the column.
    select * from X
    where x.a <> getTagValue(X.b,'elementZ')
    Check out the DBMS_LOB package.

  • Performance when using bind variables

    I'm trying to show myself that bind variables improve performance (I believe it, I just want to see it).
    I've created a simple table of 100,000 records each row a single column of type integer. I populate it with a number between 1 and 100,000
    Now, with a JAVA program I delete 2,000 of the records by performing a loop and using the loop counter in my where predicate.
    My first JAVA program runs without using bind variables as follows:
    loop
    stmt.executeUpdate("delete from nobind_test where id = " + i);
    end loop
    My second JAVA program uses bind variables as follows:
    pstmt = conn.prepareStatement("delete from bind_test where id = ?");
    loop
    pstmt.setString(1, String.valueof(i));
    rs = pstmt.executeQuery();
    end loop;
    Monitoring of v$SQL shows that program one doesn't use bind variables, and program two does use bind variables.
    The trouble is that the program that does not use bind variables runs faster than the bind variable program.
    Can anyone tell me why this would be? Is my test too simple?
    Thanks.

    [email protected] wrote:
    I'm trying to show myself that bind variables improve performance (I believe it, I just want to see it).
    I've created a simple table of 100,000 records each row a single column of type integer. I populate it with a number between 1 and 100,000
    Now, with a JAVA program I delete 2,000 of the records by performing a loop and using the loop counter in my where predicate.
    Monitoring of v$SQL shows that program one doesn't use bind variables, and program two does use bind variables.
    The trouble is that the program that does not use bind variables runs faster than the bind variable program.
    Can anyone tell me why this would be? Is my test too simple?
    The point is that you have to find out where your test is spending most of the time.
    If you've just populated a table with 100,000 records and then start to delete randomly 2,000 of them, the database has to perform a full table scan for each of the records to be deleted.
    So probably most of the time is spent scanning the table over and over again, although most of blocks might already be in your database buffer cache.
    The difference between the hard parse and the soft parse of such a simple statement might be negligible compared to effort it takes to fulfill each delete execution.
    You might want to change the setup of your test: Add a primary key constraint to your test table and delete the rows using this primary key as predicate. Then the time it takes to locate the row to delete should be negligible compared to the hard parse / soft parse difference.
    You probably need to increase your iteration count because deleting 2,000 records this way probably takes too short and introduces measuring issues. Try to delete more rows, then you should be able to spot a significant and constant difference between the two approaches.
    In order to prevent any performance issues from a potentially degenerated index due to numerous DML activities, you could also just change your test case to query for a particular column of the row corresponding to your predicate rather than deleting it.
    Regards,
    Randolf
    Oracle related stuff blog:
    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/

  • Unable to view purchased tv programmes: message authorise your computer.  How do I do that?  I used to watch same programmes using same laptop.

    I am unable to watch purchased and downloaded TV programmes on my laptop, I get the message: "Authourise your computer" How do I do that? I used to watch the same programmes on that same laptop.

    There's a couple of ways to get through to the authorisation controls in the 11.0.x versions.
    The control is still in the Store menu, but first (if you're using iTunes versions 11.0.x) you might need to bring up the menu bar to see the Store menu.
    If you're using 11.0.x, click on the wee boxy icon up in the top-left corner of your iTunes to see the "Show Menu Bar" control, as per the following screenshot:
    Then you'll find the control in the Store menu:
    Alternatively, if you don't want to bring up the menu bar, it's still possible to get into the authorise controls via nested menus accessible from the wee boxy icon. Here's a screenshot of where to find them:

  • HT201328 i had brough a used set iphone 4s from japan, how to unlock so that i can use in malaysia

    i had brough a used set iphone 4s from japan, how to unlock so that i can use in malaysia.
    PLs help i u have any ideal /way to solve. Thanks.

    if the carrier it's locked to will let you then yes
    otherwise no
    they are the only ones who have the power

  • How to use sql "IN" operator with named bind variable in where clause ?

    Can one bind variable be used for the "IN" list ('1','2','3') ?

    rob,
    No worries. Glad it helped. Glad to see also that you're doing things right and trying to use bind variables ;)
    May I suggest adding "SOLVED" to the title of the original post?
    Best,
    John

  • Query don't use the right index when using bind variables

    Hi people !
    I need some help because I have an issue with a query that don t use the right Indexes as it should
    First of all, I have mainly three tables :
    ORDER : Table that contains description for each Order (approximately 1 000 000 Records)
    ORDER_MVTS : Table that contains the tasks made (called movements) to set up each Orders
    with quantity of packages prepared for each product (approximately 10 000 000 Records)
    PRODUCT : Tables that contains the products (approximately 50 000 Records)
    When I launch the query with hard coded values, it brings back response very fast
    because it uses the right index (ORDER_DHR_VALID) which represent the date and hour of the order
    (with format 'DD/MM/YYYY HH24:MI:SS'). The selectivity for this index is good.
    NB 1: I have to use the trick " >= Trunc(date) and < trunc(date) +1 " to filter on a simple date because
    the index contains hour and minutes (I know it wasn't probably a bright idea at conception time).
    NB 2: The index on ORDER_MVTS.PRODUCT_CODE is'nt discriminating enough because there is'nt enough different products.
    It's the same for index on CUSTOMER_CODE and on MVT_TYPE so only the index on ORDER.DHR_VALID is good.
    Here is the correct explain plan when I execute the query with hard coded values :
    SELECT SUM(ORDER_MVTS.NB_PACKAGE)
    FROM ORDER_MVTS, PRODUCT, ORDER
    WHERE ORDER.DHR_VALID >= TRUNC(to_date('14/11/2008 10:04:56','DD/MM/YYYY HH24:MI:SS'))
    AND ORDER.DHR_VALID < TRUNC(to_date('14/11/2008 10:04:56','DD/MM/YYYY HH24:MI:SS')) + 1
    AND ORDER_MVTS.MVT_TYPE = 'DELIVERY'
    AND PRODUCT.CODE = ORDER_MVTS.PRODUCT_CODE
    AND ORDER_MVTS.ORDER_CODE = ORDER.CODE
    AND ORDER.CUSTOMER_CODE = 'ADIDAS'
    AND PRODUCT.CODE = 1234
    Rows Row Source Operation
    1 SORT AGGREGATE
    2 NESTED LOOPS
    4 NESTED LOOPS
    2 INDEX UNIQUE SCAN (object id 378548) --> PRODUCT_PK
    4 TABLE ACCESS BY INDEX ROWID ORDER
    777 INDEX RANGE SCAN (object id 378119) --> ORDER_DHR_VALID
    2 TABLE ACCESS BY INDEX ROWID ORDER_MVTS
    30 INDEX RANGE SCAN (object id 377784) --> ORDER_MVTS_ORDER_FK
    Now the problem is when the query is used in a Cursor with bind variables.
    It seems like Oracle don't use index on ORDER.DHR_VALID because he can't figure out that he have
    to actually filter on a short period of time (only one day).
    So Oracle uses the index on ORDER_MVTS.PRODUCT_CODE which is'nt a bright idea (it takes 10 secondes instead of just one)
    Here is the bad explain plan :
    Rows Row Source Operation
    1 SORT AGGREGATE
    2 NESTED LOOPS
    722 NESTED LOOPS
    2 INDEX UNIQUE SCAN (object id 378548) --> PRODUCT_PK
    722 TABLE ACCESS BY INDEX ROWID ORDER_MVTS
    1790 INDEX RANGE SCAN (object id 377777) --> ORDER_MVTS_PRODUCT_FK
    2 TABLE ACCESS BY INDEX ROWID ORDER
    1442 INDEX UNIQUE SCAN (object id 378439) --> ORDER_PK
    Now I have found two solutions to this problem :
    1) using a Hint to force the use of index on ORDER.DHR_VALID (with /*+ INDEX(ORDER ORDER_DHR_VALID) */ )
    2) Using Dynamic SQL and keeping the date hard coded (but not the other values except mvt_type)
    For example :
    QUERY :=
    'SELECT SUM(ORDER_MVTS.NB_PACKAGE)
    FROM ORDER_MVTS, PRODUCT, ORDER
    WHERE ORDER.DHR_VALID >= TRUNC(TO_DATE('''||To_char(P_DTE_VAL,'DD/MM/YYYY')||''',''DD/MM/YYYY'')) '||
    AND ORDER.DHR_VALID < TRUNC(TO_DATE('''||To_char(P_DTE_VAL,'DD/MM/YYYY')||''',''DD/MM/YYYY'')) + 1 '||
    AND ORDER_MVTS.MVT_TYPE = 'DELIVERY'
    AND PRODUCT.CODE = ORDER_MVTS.PRODUCT_CODE
    AND ORDER_MVTS.ORDER_CODE = ORDER.CODE
    AND ORDER.CUSTOMER_CODE = :CUSTOMER
    AND PRODUCT.CODE = :CODE ';
    These two solutions work but Number 1 is bad in theory because it uses a Hint
    and Number 2 may be difficult to code.
    So my question is : Does someone knows another solution to force the use of index ORDER_DHR_VALID that can be simple and reliable.
    Thank you very much for support
    Edited by: remaï on Apr 1, 2009 4:08 PM

    What version of oracle you have? CBO work is different in 9i and 10g.
    Usually cost based optimizer do not want to use index for >< condition with binding variables because optimizer can not use statistic to determine selectivity, and by default selectivity of <> operators is low.
    (As I remember '>' selectivity by default is 5%, you have two conditions > and <, therefore resulting selectivity will be 0.05*0.05=0.0025 as two independent events, but selectivity of other conditions
    ORDER_MVTS.MVT_TYPE = 'DELIVERY' or ORDER.CUSTOMER_CODE = 'ADIDAS' looks much better for CBO)
    The best solution I see is do not use binding variables. Actually your query looks as searching query, which executes not so often, therefore you will not have perfomance win along of skipping execution plan creation.
    Edited by: JustasVred on Apr 1, 2009 10:10 AM

  • How to use bind variables in this procedure

    Hi Experts,
    How to use bind variables in this procedure for static queries.
    PROCEDURE DELETE_MER_PROC (M_id IN NUMBER)
    IS
    BEGIN
    V_date DATE;
    SELECT PD_DATE INTO v_date FROM PD_MAINTAIN;
        DELETE FROM MER_CLEAR
        WHERE MER_DT < v_date
        AND ID = M_ID;
    COMMIT;
    END;   
    How to use  v_date and m_id as bind variables in this procedure to avoid hard parsing.
    Please help me.
    Thanks.

    976208 wrote:
    How to use  v_date and m_id as bind variables in this procedure to avoid hard parsing.
    You cannot avoid hard parsing - as the 1st time a SQL statement (like the SELECT or DELETE statements in your code) is encountered, it does not reside in the server's Shared Pool, and needs to be added into the pool via a hard parse.
    Bind variables does not prevent hard parsing. Hard parsing happens when the SQL statement (with or without bind variables) is a brand new statement encountered by the server.
    Bind variables enables the same SQL cursor to be reused, by simply changing the bind variable value.
    Not using bind variables means that each SQL statement is unique and not shareable - as the value is hardcoded into the statement and cannot be changed via a bind value. This typically means LOTS of different SQL statements (where the only difference is the changed value in the statement) are created - with each statement being a new statement not seen before in the Shared Pool and needing to be hard parsed.
    One does not design one's code not to be hard parsed. There ALWAYS will be a hard parse in order to get a SQL statement into the Shared Pool. One designs one's code to REUSE cursors in the Shared Pool.

  • PL/SQL speed issues when using a variable

    I have a very strange issue that is causing problems.
    I am running Golden connecting to a 11g database.
    I created a procedure to insert records into a table based on a query. The source query includes variables that I have populated prior to the insert statement. The problem is that if I use the variable for one very specific where statement, the statement goes from running in less than 1 second, to running in 15 minutes. It gets even more strange though. Not only does a 2nd variable not cause any problems, the exact same variable in the same statement works fine.
    This procedure takes 15 minutes to run.
    declare
        v_start_period date;
        v_end_period date;
    begin
        select add_months(trunc(sysdate,'mm'), -1), trunc(sysdate,'mm')
        into v_start_period, v_end_period
        from dual;
        insert into RESULTS_TABLE
                (first_audit_date, last_audit_date,
                data_column1, data_column2, data_column3)
            select
                a.first_audit_date, a.last_audit_date,
                b.data_column1, b.data_column2, b.data_column3
            from
                SOURCE_TABLE_1 b,
                (select marker_id, min(action_time) as first_audit_date, max(action_time) as last_audit_date
                    from SOURCE_TABLE_2
                    where action_time >= v_start_period and action_time < v_end_period
                    group by marker_id) a
            where b.marker_id = a.marker_id
                and exists (select 1 from SOURCE_TABLE_2
                    where marker_id = b.marker_id
                    and action_time >= v_start_period and action_time < v_end_period
                    and action_type = 'XYZ');
        commit;
    end;This procedure runs in less than 1 second, yet returns the exact same results.
    declare
        v_start_period date;
        v_end_period date;
    begin
        select add_months(trunc(sysdate,'mm'), -1), trunc(sysdate,'mm')
        into v_start_period, v_end_period
        from dual;
        insert into RESULTS_TABLE
                (first_audit_date, last_audit_date,
                data_column1, data_column2, data_column3)
            select
                a.first_audit_date, a.last_audit_date,
                b.data_column1, b.data_column2, b.data_column3
            from
                SOURCE_TABLE_1 b,
                (select marker_id, min(action_time) as first_audit_date, max(action_time) as last_audit_date
                    from SOURCE_TABLE_2
                    where action_time >= v_start_period and action_time < trunc(sysdate,'mm')
                    group by marker_id) a
            where b.marker_id = a.marker_id
                and exists (select 1 from SOURCE_TABLE_2
                    where marker_id = b.marker_id
                    and action_time >= v_start_period and action_time < v_end_period
                    and action_type = 'XYZ');
        commit;
    end;The only difference between the two is where I replace the first v_end_period variable with trunc(sysdate,'mm').
    I've been googling for possible solutions and keep running into something called "parameter sniffing". It appears to be a SQL Server issue though, as I cannot find solutions for Oracle. I tried nesting the insert statement inside it's own procedure with new variables populated by the old variables, but nothing changed.
    Edited by: user_7000017 on Jan 8, 2013 9:45 AM
    Edited by: user_7000017 on Jan 8, 2013 9:52 AM Put the code in code tags.

    You are not describing procedures. You are listing anonymous PL/SQL blocks.
    As for the code - this approach to assigning PL/SQL variables are highly questionable. As the functions are native PL/SQL functions too, there is no need to parse and execute a SQL cursor, do context switching, in order to assign values to PL/SQL variables.
    This is wrong:
    select
        add_months(trunc(sysdate,'mm'), -1), trunc(sysdate,'mm')
        into v_start_period, v_end_period
    from dual;This is correct:
    v_start_period := add_months(trunc(sysdate,'mm'), -1);
    v_end_period := trunc(sysdate,'mm');Just as you would not use +"select 1 into plVariable from dual;+", instead of a standard variable assignment statement like +"plVariable := 1;"+.
    As for the performance/execution difference. Does not make sense. I suggest simplifying the code in order to isolate the problem. For example, take that in-line SQL (aliased as a in the main SQL) and create a testcase where it uses the function, versus a PL/SQL bind variable with the same data type and value as that returned by the function. Examine and compare the execution plans.
    Increase complexity (if no error) and repeat. Until the error is isolated.
    The 1st step in any troubleshooting, is IDENTIFYING the problem. Without knowing what the problem is, how can one fix it?

  • How to use bind variables in desc flexfields in OA pages?

    Use case: In EBS 11i we have a DFF on po_vendors and one of the segment has a table validated value set that is referencing a forms variable i.e :VENDORS.VENDOR_ID. Now after upgrading to R12 this won't work as the vendor pages moved to OA framework. Is there any way to reference the vendor_id column in OA framework?

    976208 wrote:
    How to use  v_date and m_id as bind variables in this procedure to avoid hard parsing.
    You cannot avoid hard parsing - as the 1st time a SQL statement (like the SELECT or DELETE statements in your code) is encountered, it does not reside in the server's Shared Pool, and needs to be added into the pool via a hard parse.
    Bind variables does not prevent hard parsing. Hard parsing happens when the SQL statement (with or without bind variables) is a brand new statement encountered by the server.
    Bind variables enables the same SQL cursor to be reused, by simply changing the bind variable value.
    Not using bind variables means that each SQL statement is unique and not shareable - as the value is hardcoded into the statement and cannot be changed via a bind value. This typically means LOTS of different SQL statements (where the only difference is the changed value in the statement) are created - with each statement being a new statement not seen before in the Shared Pool and needing to be hard parsed.
    One does not design one's code not to be hard parsed. There ALWAYS will be a hard parse in order to get a SQL statement into the Shared Pool. One designs one's code to REUSE cursors in the Shared Pool.

  • PL SQL using bind variables

    I am very much new to PL/SQL to pardon my neivte.
    I would like to have some PL/SQL code that uses
    bind variables for insert statements and than issue
    commit's after a specific number of inserts.
    For example.
    desc xxx.taba
    a number
    b varchar2(20)
    I want to do someting like this:
    declare
    l_start number default dbms_utility.get_time;
    create or replace procedure insert_xxx(commit in number)
    begin
    for i in 1 .. 1000
    loop
    :a:=i;
    :b:=dbms_random.string('P', 20);
    intert into xxx :a :b;
    if i/mod=(number passed into to proc)
    commit
    end loop;
    dbms_output.put_line
    (round((dbms_utility.get_time-l_start)/100, 2) ||
    ' Seconds...' );
    end;
    end;
    Can somebody help me out with the correct syntax
    as I know it's incorrect.

    A few remarks
    - procedures don't have hidden variables defined in the outside world, as procedures are black boxes, which communicate only with the outside world
    using the actual parameter list
    - procedures don't commit. The caller should always commit!!!
    - procedures don't commit every <n> records. A commit finalizes a logical unit of work, not a technical one.
    If I would have been your supervisor and you would have presented this code, I would have fired you on the spot.
    Committing inside a loop also increases the likelihood of ora-1555 errors.
    So: remove the current parameter
    a and b should be formal parameters of the procedure
    Inside the procedure it should read
    insert into xxx values (a, b)
    Sybrand Bakker
    Senior Oracle DBA

  • Correct Way to use Bind variables when using an MS SQL server backend

    Hey,
    I'm having some trouble figuring out how to use bind variables in a view when using an MS SQL server backend. For some reason when I'm using an ApplicationModule that has a MS SQL Server JDBC library loaded and I attempt to click OK when editing the following query:
    SELECT kit_status, component_id
    FROM numbered_inv
    WHERE trialname = :1
    I get an error stating that "SQL Query Error Message Incorrect Syntax near ':'". Is JDeveloper compatible with SQL server for bind variables because this query runs just fine if I replace the :1 with a say "Testing testing".
    Thanks in advance
    Edited by: NullCheck on Dec 15, 2010 2:06 PM

    Gotcha. I had tried %1% %0% ?0 and ?1 and various other permutations none of which worked. Got it working by setting the query to:
    SELECT kit_status, component_id
    FROM numbered_inv
    WHERE trialname = ?
    ORDER BY kit_status
    And then adding a bind variable with "Bind position" set to 0.
    Thanks again Shay!

  • How to use bind variables in the following query

    CREATE OR REPLACE PROCEDURE MMDB.test IS
    sel_qtn VARCHAR2 (10);
    CURSOR PT_QUANTITY IS select * from mmdb.product_tree WHERE QUANTITY_CHECK ='E'
    AND run_id = 100
    a PT_QUANTITY%ROWTYPE;
    BEGIN
    FOR i IN PT_QUANTITY
    loop
    sel_qtn := i.quanttity-1;
    While sel_qtn>=1
    loop
    insert into mmdb.product_tree (BILLING_ACCOUNT_NO ,S_CODE) values (i.BILLING_ACCOUNT_NO ,i.S_CODE||'E');
    sel_qtn :=sel_qtn -1;
    End loop;
    commit;
    end;

    Don't duplicate threads: How to use bind variables in the following query

  • How to use bind variables in JSP applications ?

    Hi,
    How will i use bind variables in jsp applicaions ?
    Any links that give a clear understanding ?
    TIA,
    jj

    Hi,
    Try this link, this may help you.
    http://www.oracle-base.com/articles/misc/WebScriptingForOracle.php
    Thanks

  • ORA-01722: invalid number caused with SQL using bind variable

    Hi,
    Im am hoping that someone can help me resolve a problem thats only just services and is being experienced on quite a few clients.
    our application uses C++ exes and makes OCI calls to the database.
    what has happened in the last week or so, there has been quite a few invalid number errors being received on a prod server but strangly enough we cannot reproduct the error on our UAT system.
    The sql is using bind variables and the information in the trace file shows that a number is being used for the bind variable, here is an extract from one trace file:
    for some schemas, the bind variable value is some currupted value, i think:
    oacdty=01 mxl=32(21) mxlc=00 mal=00 scl=00 pre=00
    oacflg=03 fl2=1206001 frm=01 csi=178 siz=32 off=0
    kxsbbbfp=9a8d62b8 bln=32 avl=03 flg=05
    value="Â*d"
    but on another schema, the value used is:
    Bind#0
    oacdty=01 mxl=32(32) mxlc=00 mal=00 scl=00 pre=00
    oacflg=03 fl2=1206001 frm=01 csi=178 siz=32 off=0
    kxsbbbfp=c5f92718 bln=32 avl=04 flg=05
    value="2101"
    however both produce invalid number errors.
    I am relatively inexperienced as a DBA so would appreciate as much help as i can get.

    Could you post your sql statement that is being run.
    Also post the query plan from your uat system and the one from your production system (They are likely to be different)
    You can export the stats from your production system and run them in your uat system. If you do this, then the execution plans should be the same on both systems(dbms_stats) and if you have the same data you should run into the same problem on uat as in production.
    The root cause of this type of problem is having a column in a table which holds values which are of different datatypes. Typically there is a condition in the where clause which indicates that for example only numeric columns should be retrieved from the column which holds multiple data types. However since the optimizer is free to rewrite the query any way it sees fit, (It does not necessarily execute in the order the sql statement is written) it does not filter this data first, and therefore you hit non-numeric data and run into the invalid number error.
    You can use little techniques like using an inline view with rownum in the column list, to perform the first filter. This ensures that the inline view is executed on its own, rather than being merged (materialized) with the rest of the query.
    A quick temporary solution is to use a comparison like to_char(column_name) = variable
    Make sure your comparisons are correct and it doesn't negatively impact on performance of the query

Maybe you are looking for