Updating or selecting every second row

[philippefbertrand]

My ultimate goal is to select every second row (by primary key order) without any additional columns in the result set. I want to do any manipulation in dbisql but the final SELECT will be in a MobiLink script. I need to repeat this on multiple tables and can make no assumptions about the pkey values or type. I can add a column to each table if needed.

Can it be done in a single select statement (without adding a column)? (I know SELECT * FROM Foo WHERE mod(number(*),2) = 0 ORDER BY num does not work)

If I add a column to all the tables, can a single update be written to update every second column?

If the rows are inserted in primary key order, chances are adding a column with default autoincrement should number them in order and then I select every odd or even value but there is no guarantee of the order.

I'm working with SQL Anywhere 11 in this particular problem.

[philippefbertrand]

Yes, I could have just walked down the hallway but when I first pondered this, the first answers I got weren't satisfactory. Besides, there aren't enough question here - perhaps its just because our docs are so great!

[philippefbertrand · Accepted solution]

After messing up the syntax a few times, I got a single SELECT to work:

SELECT Ba.num, Ba.MyID FROM
   (SELECT num, MyID, row_number() OVER( ORDER BY num ) From Foo) AS Ba(num,MyID,bar)
WHERE mod(Ba.bar,2) = 0

Still curious if a single UPDATE could be written to update every second row.

[MarkCulp · Accepted solution]

The update part is simple once you have the query that selects the PKs of the rows to be updated - the general form is:

UPDATE T SET ... WHERE T.pk in ( <your-query-that-selects-the-pks-to-be-updated> )

[philippefbertrand · Accepted solution]

Yes, now that I figured out how to control the select list, it is obvious.