4

Static Cursor doesn't allow to modify the data because it's read only, and when executed with "Where current of", it returns error as expected. So far so good. But i was surprised to find that static cursor allow to modify data with variable like this. 

DECLARE @nome varchar(100), @salario int,@idemp int
DECLARE contact_cursor  CURSOR STATIC FOR
SELECT empno,ename, sal FROM emp



OPEN contact_cursor;



FETCH NEXT from contact_cursor into @idemp,@nome, @salario
WHILE @@FETCH_STATUS=0
BEGIN



If @salario < 5000
    Update Emp
    Set Sal = Sal * 1.1
    where empno=@idemp --No error and do the update
    --Where current of contact_cursor; --gives error

print @nome+' '+cast(@salario as varchar(100));




FETCH NEXT from contact_cursor into @idemp,@nome, @salario
END
CLOSE contact_cursor;
DEALLOCATE contact_cursor;
The question is: Which is the difference between using "where current" and the variable extracted with the cursor in this update?

Improve this question

2 Answers 2

Reset to default
5

The main difference seems to be how each approach finds the row to be updated. The STATIC Cursor copies the full result set to a hidden temporary table first (hence why it is read-only), so it would seem to be less efficient to then have to re-query the main table for each UPDATE. However, the Positioned Update seems to have quite a bit more in Logical Reads and operations. One advantage of the Positioned Update, however, is noted in the MSDN page for UPDATE:

CURRENT OF

Specifies that the update is performed at the current position of the specified cursor.

A positioned update using a WHERE CURRENT OF clause updates the single row at the current position of the cursor. This can be more accurate than a searched update that uses a WHERE clause to qualify the rows to be updated. A searched update modifies multiple rows when the search condition does not uniquely identify a single row.

Test Setup

SET NOCOUNT ON;
-- DROP TABLE ##CursorTest;
CREATE TABLE ##CursorTest ([ID] INT IDENTITY(1, 1) NOT NULL PRIMARY KEY,
                           [Val] INT NOT NULL);
INSERT INTO ##CursorTest ([Val]) VALUES (1), (1), (1), (1);

Updateable CURSOR and WHERE CURRENT OF

UPDATE ##CursorTest SET [Val] = 1;
SELECT * FROM ##CursorTest;

SET STATISTICS IO ON;
DECLARE curTest CURSOR TYPE_WARNING
  LOCAL
  FORWARD_ONLY
  KEYSET -- removing only reduces logical reads by 4
  SCROLL_LOCKS
  --OPTIMISTIC 
FOR
  SELECT [ID] FROM ##CursorTest WHERE [Val] < 5
  FOR UPDATE OF [Val];

DECLARE @ID INT;
OPEN curTest;

FETCH NEXT
FROM  curTest
INTO  @ID;

WHILE (@@FETCH_STATUS = 0)
BEGIN
  UPDATE tmp
  SET    tmp.[Val] = tmp.[Val] + 2
  FROM   ##CursorTest tmp
  WHERE CURRENT OF curTest;

  FETCH NEXT
  FROM  curTest
  INTO  @ID;
END;

CLOSE curTest;
DEALLOCATE curTest;
SET STATISTICS IO OFF;

SELECT * FROM ##CursorTest;

Results:

Table 'Worktable'. Scan count 0, logical reads 8
Table '##CursorTest'. Scan count 1, logical reads 2
Table '##CursorTest'. Scan count 1, logical reads 2
Table 'Worktable'. Scan count 1, logical reads 2
Table '##CursorTest'. Scan count 0, logical reads 2
Table 'Worktable'. Scan count 0, logical reads 2
Table '##CursorTest'. Scan count 1, logical reads 2
Table 'Worktable'. Scan count 1, logical reads 2
Table '##CursorTest'. Scan count 0, logical reads 2
Table 'Worktable'. Scan count 0, logical reads 2
Table '##CursorTest'. Scan count 1, logical reads 2
Table 'Worktable'. Scan count 1, logical reads 2
Table '##CursorTest'. Scan count 0, logical reads 2
Table 'Worktable'. Scan count 0, logical reads 2
Table '##CursorTest'. Scan count 1, logical reads 2
Table 'Worktable'. Scan count 1, logical reads 2
Table '##CursorTest'. Scan count 0, logical reads 2
Table 'Worktable'. Scan count 0, logical reads 2
Table '##CursorTest'. Scan count 0, logical reads 0
Table 'Worktable'. Scan count 1, logical reads 2

Removing the KEYSET option did reduce the logical reads by 4 (I believe), but that might not be a savings on a more complicated query, possibly with JOINs.

Also, switching SCROLL_LOCKS to be OPTIMISTIC increased the Logical Reads.

STATIC Cursor and standard UPDATE

UPDATE ##CursorTest SET [Val] = 1;
SELECT * FROM ##CursorTest;

SET STATISTICS IO ON;
DECLARE curTest CURSOR TYPE_WARNING
  LOCAL
  FORWARD_ONLY
  STATIC
  OPTIMISTIC 
FOR
  SELECT [ID] FROM ##CursorTest WHERE [Val] < 5;

DECLARE @ID INT;

OPEN curTest;

FETCH NEXT
FROM  curTest
INTO  @ID;

WHILE (@@FETCH_STATUS = 0)
BEGIN
  UPDATE tmp
  SET    tmp.[Val] = tmp.[Val] + 2
  FROM   ##CursorTest tmp
  WHERE  tmp.[ID] = @ID;

  FETCH NEXT
  FROM  curTest
  INTO  @ID;
END;

CLOSE curTest;
DEALLOCATE curTest;
SET STATISTICS IO OFF;

SELECT * FROM ##CursorTest;

Results:

Table 'Worktable'. Scan count 0, logical reads 8
Table '##CursorTest'. Scan count 1, logical reads 2
Table 'Worktable'. Scan count 0, logical reads 2
Table '##CursorTest'. Scan count 0, logical reads 2
Table 'Worktable'. Scan count 0, logical reads 2
Table '##CursorTest'. Scan count 0, logical reads 2
Table 'Worktable'. Scan count 0, logical reads 2
Table '##CursorTest'. Scan count 0, logical reads 2
Table 'Worktable'. Scan count 0, logical reads 2
Table '##CursorTest'. Scan count 0, logical reads 2
Table 'Worktable'. Scan count 0, logical reads 2

These simple tests seem to show the STATIC Cursor and regular UPDATE being the better option, and a more complicated query for the Cursor might be an even bigger difference (assuming you are able to update based on the Clustered Key of the target table).

But, if you have a situation where you can't narrow down to a individual row / have no Key value to use, then the Positioned Update would be quite handy.

Improve this answer
3

To better understand this update behavior, we may need to look at UPDATE syntax, it states:

CURRENT OF

Specifies that the update is performed at the current position of the specified cursor.

A positioned update using a WHERE CURRENT OF clause updates the single row at the current position of the cursor. This can be more accurate than a searched update that uses a WHERE clause to qualify the rows to be updated. A searched update modifies multiple rows when the search condition does not uniquely identify a single row.

So this means if you use "CURRENT OF", it is an update on the specified cursor, and with static cursor (which is read-only), we will get the error you posted.

However, when you are not using "CURRENT OF", you are doing the update on the underlying table directly, which in such case has nothing to do with the specified cursor, thus it does not matter whether the cursor is ReadOnly or NOT.

Also according to static cursor definition on MSDN

STATIC

Defines a cursor that makes a temporary copy of the data to be used by the cursor. All requests to the cursor are answered from this temporary table in tempdb; therefore, modifications made to base tables are not reflected in the data returned by fetches made to this cursor, and this cursor does not allow modifications.

So in this context, when you make any updates to the underlying table of this static cursor, the changes will not be reflected back to the cursor (i.e. the temporary table that represents the cursor).

If I can make a quick logic summary (in my own understanding)

When you create a static cursor, a tempory table in [TempDB] is created and this temp table is read only, meaning you cannot use CURRENT OF to refer to this temp table and do any updates on it.

But you can do whatever you want to do to the underlying table (for this static cursor)

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.