I am new to Oracle PL/SQL and am trying to adjust from SQL Server with C# to a Oracle custom web application using PL/SQL for multiple layers of the application.
I currectly am having issues with rolling back a transaction. Since I am using to writing reusable code, I have written my PL/SQL code using packages and methods within those classes. My update procedure first completes validation and if validated successfully, calls a succession of various package methods to complete the save. Unfortunately, my rollback in the EXCEPTION portion of the update procedure does not rollback everything when the Rollback function is called. I'm at a loss as to why it is doing this. My basic code (although not exact due to legal issues) is as follows:
PROCEDURE SaveApplicationData(
variableName IN VARCHAR2 DEFAULT NULL,
--...
seq_id OUT INT )
AS
BEGIN
SET TRANSACTION NAME 'Transaction Name';
--Save initial program record
SaveNewRecord(variableName, seq_id);
IF (seq_id != 0) THEN
--If saved successfully, seq_id represents record ID
package_class.secondarySaveMethod(variableName, seq_id);
second_package_class.anotherSaveMethod(variableName, seq_id);
END IF;
COMMIT;
htp.p('Sequence ID: ' || seq_id);
htp.p('Saved the record"' || programName || '" successfully!');
EXCEPTION
WHEN OTHERS THEN
utilityPackage.rollbacktransaction;
END SaveApplicationData;
The utilityPackage.rollbacktransaction includes a ROLLBACK as well as the custom error exception handling package used by our organization.
Essentially, it will roll back the section that causes the error, but as soon as it rolls the section back, it continues with the rest of the transaction (and does not roll back previously executed blocks of code).
Please let me know if this does not make sense - and thank you in advance for your help!
Coming from a SQL Server environment before Oracle I can understand the confusion. Oracle does not use a BEGIN TRANSACTION. Instead, a transaction is implicitly started for you.
Thus, I believe in your case SET TRANSACTION NAME is not what you are looking to do, see
SET TRANSACTION.
I recommend removing the rollback code from your package and placing it in C#. It should be the responsibility of the caller to commit. Use a transaction in C# to guarantee the the transaction is committed upon successful execution of the package(s).
Ideally, your package structure should look more like this.
declare
ex_custom EXCEPTION;
PRAGMA EXCEPTION_INIT( ex_custom, -20001 );
begin
--Save initial program record
SaveNewRecord(variableName, seq_id);
IF (seq_id != 0) THEN
--If saved successfully, seq_id represents record ID
package_class.secondarySaveMethod(variableName, seq_id);
second_package_class.anotherSaveMethod(variableName, seq_id);
ELSE
-- seq_id invalid throw an exception
RAISE_APPLICATION_ERROR(-20001,'Custom error')
END IF;
htp.p('Sequence ID: ' || seq_id);
htp.p('Saved the record"' || programName || '" successfully!');
EXCEPTION
WHEN ex_custom THEN
-- if needed we log it
utility.log_exception;
-- Raise it for the client to handle
raise;
END SaveApplicationData;
Related
In SQL Developer, when running some PL/SQL, when the procedure is completed, the message 'PL/SQL procedure successfully completed.' is returned.
The PL/SQL that is run may return error messages to the user if the operation could not be completed for any reason through DBMS_OUTPUT.PUT_LINE, however, the user will also see 'PL/SQL procedure successfully completed.', which could be misleading (especially if the Script Output window is small enough that the DBMS_OUTPUT is not visible).
Is there any way to have the DBMS_OUTPUT return what it should whilst also having the script not return 'PL/SQL procedure successfully completed.'?
If not, are there any alternatives in SQL Developer that I may be unaware of to provide instant personalised feedback to the user?
declare
testex exception;
begin
if 1=1 then
raise testex;
end if;
exception when testex then
dbms_output.put_line('Error msg');
end;
Below code works in my end. Did you try to run your code like below?
Copied text from a website to explain the SET FEEDBACK OFF command.
Source link: https://docs.oracle.com/cd/B19306_01/server.102/b14357/ch12040.htm
SET FEED[BACK] {6 | n | ON | OFF} -
Displays the number of records returned by a script when a script selects at least n records.
ON or OFF turns this display on or off. Turning feedback ON sets n to 1. Setting feedback to zero is equivalent to turning it OFF.
SET FEEDBACK OFF also turns off the statement confirmation messages such as 'Table created' and 'PL/SQL procedure successfully completed' that are displayed after successful SQL or PL/SQL statements.
Add a RAISE statement in the error handler to re-raise the exception so that any outer handler can deal with it:
declare
testex exception;
begin
if 1=1 then
raise testex;
end if;
exception when testex then
dbms_output.put_line('Error msg');
RAISE; -- re-raise the exception to an outer handler
end;
Best of luck.
You have coded it explicitly to complete successfully in the event of a testex exception. The code says: if this happens, then print a message and end. If you want it to actually fail then it needs to raise an exception.
I would just use something like this:
begin
if 1=1 then
raise_application_error(-20000, 'Bananas are not available on a Tuesday');
end if;
end;
/
begin
*
ERROR at line 1:
ORA-20000: Bananas are not available on a Tuesday
ORA-06512: at line 3
This will cause an actual error, rather than just printing out a message that happens to talk about an error, allowing you to control deployments and build scripts etc.
It will also roll back any uncommitted transactions within the block, which your current approach does not, and it will show the actual line number.
Hello i've this function to update my Access DB using TUniQuery :
var
Res:Boolean;
begin
Res:=false;
try
with MyQuery do
begin
Active := false;
SQL.Clear;
SQL.Add('Update MYTABLE');
SQL.Add('set username='+QuotedStr(NewUserName));
SQL.Add(',password='+QuotedStr(NewPassword));
SQL.Add('where username='+QuotedStr(ACurrentUserName));
ExecSQL;
Res:=true;
end;
except
Res:=False;
end ;
Result:=Res;
end;
Is the use of Try ... except enough to KNOW when " ExecSQL " succeeds or fails ?
or is there any other better approach ?
thank you
You may want to consider your update succeeded if no exception is raised. It means the database is responsive and parsed and executed your statement without syntax errors.
In a statement like shown, you may also want to be sure that exactly one row was updated, because I assume that's your intention.
To check that, you can resort to the result of the ExecSQL method, which returns the number of rows affected by the execution of the statement. So, you may change your code to this:
begin
with MyQuery do
begin
Active := false;
SQL.Clear;
SQL.Add('Update MYTABLE');
SQL.Add('set username='+QuotedStr(NewUserName));
SQL.Add(',password='+QuotedStr(NewPassword));
SQL.Add('where username='+QuotedStr(ACurrentUserName));
Result := ExecSQL = 1; //exactly 1 row updated
end;
end;
I also changed the unconditional exception handler, since it may not be the proper site to eat any exception, and also removed the local variable to store the Result, since that really is not necessary.
After reading your added text and re-thinking your question:
Is the use of Try ... except enough to KNOW when " ExecSQL " succeeds or fails ?
You really have to change your mind about exception handling and returning a boolean from this routine. Exceptions were introduced as a whole new concept on how to address exceptional and error situations on your programs, but you're killing this whole new (and IMHO better) approach and resorting to the old way to return a value indicating success or failure.
Particularly a try/exception block that eats any exception is a bad practice, since you will be killing exceptions that may be raised for too many reasons: out of memory, network problems like connection lost to database, etc.
You have to re-think your approach and handle these exceptional or error conditions at the appropriate level on your application.
My advise is:
change this from a function to a procedure, the new contract is: it returns only if succeded, otherwise an exception is raised.
if an exception occurs let it fly out of the routine and handle that situation elsewhere
raise your own exception in case no exactly 1 row is updated
change the query to use parameters (avoiding sql injection)
The routine may look like this:
procedure TMySecurityManager.ChangeUserNameAndPassword();
begin
MyQuery.SQL.Text := 'Update MYTABLE'
+ ' set username = :NewUserName'
+ ' , password = :NewPassword'
+ ' where username = :username';
MyQuery.Params.ParamByName('NewUserName').AsString := NewUserName;
MyQuery.Params.ParamByName('NewPassword').AsString := NewPassword;
MyQuery.Params.ParamByName('username').AsString := ACurrentUserName;
if MyQuery.ExecSQL <> 1 then
raise EUpdateFailed.Create('A internal error occurred while updating the username and password');
//EUpdateFailed is a hypotetical exception class you defined.
end;
I have a series of update statements that I need to use in my Oracle package. It's rare but there may be an occasional and unavoidable user error that would result in one of the update statements throwing a "Single row sub-query returns one or more rows" Error.
I've been looking into exception handling for oracle PL/SQl and I'm a bit stuck on how and what to use to catch this exception so the package doesn't crash.
I know of the pre-built "Too Many Rows" exception clause that exists but everything I read seems to say it is used for improper insert statements.
Can I use this as my exception? Or do I need to build my own exception clause. I've never built one myself before and have only a rough idea on where to put everything needed for it.
The following code is basically how the updates are set up in this particular procedure
but for the sake of brevity I'm only using a bare bones example of how it looks.
INSERT INTO TempTable... --(Initial insert statement)
UPDATE TempTable t SET t.Row_one = (SELECT (Statement_One))
WHERE T.Row_One is NULL
UPDATE TempTable t SET t.Row_one = (SELECT (Statement_Two))
WHERE T.Row_One is NULL
UPDATE TempTable t SET t.Row_one = (SELECT (Statement_Three))
WHERE T.Row_One is NULL
-- Does the exception clause start here?
EXCEPTION
WHEN TOO_MANY_ROWS THEN
(What do I tell the Procedure to do here, what am I able to tell it to do?)
--end of updates that need the exception handling
-- more insert statements into other tables based on data from the preceding Temp Table
END;
Will this work or do I need to build a custom exception?
Thanks in advance.
First, the TOO_MANY_ROWS exception will not catch the case where your select statements return multiple rows. The TOO_MANY_ROWS exception is for ORA-01422 when you issue a SELECT .. INTO statement that returns more than one row. The exception you'll encounter in your case is ORA-01427, Single row subquery returns more than one row.
If you want to handle this specific error in your procedure, use the EXCEPTION_INIT pragma to associate an exception name with the error:
too_many_values EXCEPTION;
PRAGMA EXCEPTION_INIT(too_many_values, -1427);
Then you can reference this name in your exception handler:
EXCEPTION
WHEN TOO_MANY_VALUES THEN
{perform your handler here}
What you put in the handler depends on what your procedure does. Many times you'll want to return some sort of error code/message to your caller:
PROCEDURE my_proc(p_one VARCHAR2, p_err OUT VARCHAR2) IS
too_many_values EXCEPTION;
PRAGMA EXCEPTION_INIT(too_many_values, -1427);
BEGIN
...
EXCEPTION
WHEN TOO_MANY_VALUES THEN
p_err := 'More than one value available to assign in the update';
RAISE; -- re-raise the exception for the caller
WHEN OTHERS THEN
p_err := SQLERRM; -- return the oracle message for the unexpected error
RAISE;
END;
Another approach is to skip the specific exception handlers and return generic oracle messages in the WHEN OTHERS handler:
EXCEPTION
WHEN OTHERS THEN
p_err := SQLERRM;
END;
The advantage with the first approach is you can customize your messages to be more end-user friendly when the output from the process is fed directly back to the user. The advantage with the latter approach is there is less coding involved. Error handling is an important and often skimped on aspect of any application.
The documentation from Oracle is here.
EDIT:
If this is a package, and you want to avoid passing a long chain of error variables through a series of procedure calls, you could declare an error variable with package scope, set it when the error is encountered, and RAISE the error again.
PACKAGE BODY my_pkg is
g_err VARCHAR2(256);
PROCEDURE procx(... , p_err OUT VARCHAR2) IS...
...
proc_y(p1);
EXCEPTION
WHEN OTHERS THEN
p_err := NVL(g_err, SQLERRM);
END;
PROCEDURE proc_y(p1 VARCHAR2) IS
...
proc_z(p2);
END;
PROCEDURE proc_z(p2 VARCHAR2) IS
too_many_values EXCEPTION;
PRAGMA EXCEPTION_INIT(too_many_values, -1427);
BEGIN
....
EXCEPTION
WHEN TOO_MANY_VALUES THEN
g_err := 'More than one value available to assign in the update';
RAISE; -- re-raise the exception for the caller
END;
When the exception is raised in proc_z, it is handled and then raised again. It propagates back through proc_y (no handler there) and then gets returned to the user in proc_x. Errors not set in the global g_err get the generic Oracle error message. This avoids having to pass the initial error parameter throughout the package.
I have a bug I've been trying to track down for a few weeks now, and have pretty much isolated exactly what's going wrong. However, why this happens is beyond me.
I have a very large Oracle package (around 4,100 lines of code) and I'm calling several procedures. However, data seems to be getting lost between procedure calls.
The data that is being lost is:
dpmethodstate varchar_state_local_type;
First, I call this procedure:
PROCEDURE delivery_plan_set_state (
messages OUT ReferenceCursor,
state IN varchar_state_local_type
) AS
BEGIN
logMessage('state COUNT is: ' || state.COUNT);
dpmethodstate := state;
FOR I IN 1..dpmethodstate.COUNT LOOP
logMessage(dpmethodstate(I));
END LOOP;
logMessage('delivery_plan_set_state end - dpmethodstate count is now ' || dpmethodstate.COUNT);
OPEN messages FOR SELECT * FROM TABLE(messageQueue);
messageQueue := NULL;
END delivery_plan_set_state;
I pass in state, which is a valid array of a single string. I can verify in the logs that dpmethodstate has a COUNT of 1 when the procedure ends.
Next, I call the execute_filter procedure which looks like this:
PROCEDURE execute_filter (
--Whole bunch of OUT parameters
) AS
--About 50 different local variables being set here
BEGIN
SELECT TO_CHAR(SYSTIMESTAMP, 'HH24:MI:SS.ff') INTO TIMING FROM DUAL;
logMessage('[' || TIMING || '] execute_filter begin');
logMessage('a) dpmethodstate Count is: ' || dpmethodstate.COUNT);
--Rest of procedure
However, this time dpmethodstate.COUNT is 0. The value I set from delivery_plan_set_state has vanished!
When I look at my logs, it looks something like this:
proposed
delivery_plan_set_state end - dpmethodstate count is now 1
[21:39:48.719017] execute_filter begin
a) dpmethodstate Count is: 0
As you can see, dpmethodstate got lost between procedure calls. There's a few things to note:
Nothing else in this package is capable of setting the value for dpmethodstate besides delivery_plan_set_state. And I can see nothing else has called it.
My client side code is written in C#, and not much happens between the two procedure calls.
This happens perhaps 1 out of every 100 times, so it's very difficult to track down or debug.
First off, what's the best way to debug a problem like this? Is there any more logging I can do? Also, how does Oracle maintain state between procedure calls and is there anything that can intermittently reset this state? Any pointers would be much appreciated!
Is dpmethodstate a package global variable? I'm assuming it is, but I don't see that explicitly mentioned.
Since package global variables have session scope, are you certain that the two procedure calls are always using the same physical database connection and that nothing is using this connection in the interim? If you're using some sort of connection pool where you get a connection from the pool before each call and return the connection to the pool after the first call, it wouldn't be terribly unusual in a development environment (or low usage environment) to get the same connection 99% of the time for the second call but get a different session 1% of the time.
Can you log the SID and SERIAL# of the session where you are setting the value and where you are retrieving the value?
SELECT sid, serial#
FROM v$session
WHERE sid = sys_context( 'USERENV', 'SID' );
If those are different, you wouldn't expect the value to persist.
Beyond that, there are other ways to clear session state but they require someone to take explicit action. Calling DBMS_SESSION.RESET_PACKAGE or DBMS_SESSION.MODIFY_PACKAGE_STATE(DBMS_SESSION.REINITIALIZE) will clear out any session state set in your session. Compiling the package will do the same thing but that should throw an error warning you that your session state was discarded when you try to read it.
I'm having difficulty debugging triggers in Oracle. Currently I'm using Oracle's Sql Developer tool.
To test a trigger I write inserts or deletes in a test window but I can't see what is happening inside the trigger. I would like to step through the trigger and see what is happening as the trigger fires. Is there any way to use select statements to display variable values inside the triggers?
Firstly, don't "start from here", or more specifically, don't use triggers. Triggers are going to force switching to row-level processing if the triggers are going to fire for each row. It's better to put the logic in a stored procedure which you call. You've then got a start (where you validate inputs) and an end and a logic path all the way through. Stored procedures are a lot easier to debug as you follow one path.
Second, never test for an error you don't know how to handle. If you don't catch it, it bubbles up to the client who gets an error report saying what went wrong (error message) and where (i.e. the error/call stack). If you try to catch it, you have to know what to do with it (and if you don't know the tendency is to ignore it - which is BAD).
Finally, you can't readily see each 'layer' of a select. The explain plan will generally tell you how its going about things. v$session_longops MAY indicate what it is currently doing. The current wait event MAY give clues as to what table/block/row it is currently working on.
A rough-and-ready simple method if you must debug triggers is to use DBMS_OUTPUT.
e.g.
SQL> CREATE OR REPLACE TRIGGER mytrigger
BEFORE UPDATE ON mytable
FOR EACH ROW
...
BEGIN
DBMS_OUTPUT.put_line('mytrigger STARTING');
... do some logic ...
DBMS_OUTPUT.put_line('old=' || :OLD.mycolumn);
DBMS_OUTPUT.put_line('new=' || :NEW.mycolumn);
DBMS_OUTPUT.put_line('mytrigger FINISHED');
END;
/
SQL> SET SERVEROUT ON
SQL> UPDATE mytable SET mycolumn = mycolumn + 1;
2 rows updated.
mytrigger STARTING
old=10
new=11
mytrigger FINISHED
mytrigger STARTING
old=20
new=21
mytrigger FINISHED
Application
I use a program from Quest called TOAD available at www.quest.com/toad/toad-for-oracle.aspx.
As mentioned above, DBMS_OUTPUT is very handy. In your editor, make sure you enable the Output window.
PL/SQL works on "blocks" of code and you can catch it with an EXCEPTION keyword.
(Please forgive my formatting, not sure how to format for web)
DECLARE
C_DATE_FORMAT VARCHAR2(20) := 'DD-Mon-YYYY';
C_TIME_FORMAT VARCHAR2(20) := 'HH24:MI:SS';
C_NOT_IMPLEMENTED_CODE CONSTANT NUMBER(5) := -20200;
C_NOT_IMPLEMENTED_MESSAGE CONSTANT VARCHAR2(255) := 'Not implemented';
not_implemented EXCEPTION; -- user defined exception
BEGIN
--RAISE not_implemented; -- raise user defined exception
RAISE_APPLICATION_ERROR(C_NOT_IMPLEMENTED_CODE, C_NOT_IMPLEMENTED_MESSAGE); -- user defined exception
EXCEPTION -- exception block
WHEN not_implemented THEN -- catch not_implemented exception
DBMS_OUTPUT.PUT_LINE('Error: Not implemented');
WHEN OTHERS THEN -- catch all other exceptions
DBMS_OUTPUT.PUT_LINE('Error occured.');
DBMS_OUTPUT.PUT_LINE('Date: ' || TO_CHAR(SYSDATE, C_DATE_FORMAT));
DBMS_OUTPUT.PUT_LINE('Time: ' || TO_CHAR(SYSDATE, C_TIME_FORMAT));
DBMS_OUTPUT.PUT_LINE('Error code: ' || SQLCODE);
DBMS_OUTPUT.PUT_LINE('Error message: ' || SQLERRM); --deal with error
RAISE; -- raise to calling object
END;
Follow below steps to DEBUG trigger in Oracle PL/SQL
Open your trigger using Edit option. Set breakpoint by clicking on line number (See below).
Open Test Window as shown below.
Write Insert/Update query for your table. (Note: Your trigger could be set on update/insert or both)
Click Start debugger button.
As soon as value will be updated in your Table. Debugger will take you to trigger. Where you can continue your debugging.