Problem solve Get help with specific problems with your technologies, process and projects.

A "simple" crosstab problem

I am trying to build a query on multiple tables. I have created a view based on the multiple tables (8). I have following columns in the view -- CATEGORY, SALES_OFFICE, SALES_GROUP, STARTWEEK, ENDWEEK, DURATION and PRICE. If CATEGORY, SALES_OFFICE, SALES_GROUP, STARTWEEK, ENDWEEK and DURATION is the same for multiple records (it would occur max 4 occurrences, i.e. 4 records), then it should be displayed in the following way: CATEGORY SALESOFFICE SALESGROUP STARTWEEK ENDWEEK DURATION PRICE1 PRICE2 PRICE3 PRICE4.

Also the records should not be duplicated. I tried to generate a query to display data in the above format, but I get a cartesian product. Any thoughts on how to go ahead with it?

With the cartesian product, you were half way to a solution -- not a good solution, but a solution that could work.

Let's call your view V and let's call all those other columns you mentioned C. The following query will do what you want --

select v1.C
     , v1.PRICE, v2.PRICE, v3.PRICE, v4.PRICE
  from V as v1
left outer join V as v2 on v1.C = v2.C
left outer join V as v3 on v1.C = v3.C
left outer join V as v4 on v1.C = v4.C

The joins are LEFT OUTER in case a given value of C occurs fewer than 4 times. The problem with this 4-way join, though, and the reason it's such a poor solution, is that it will take forever to run. V isn't just a table, it's a view, containing eight tables. The optimizer has to assemble 32 tables into a working retrieval strategy, and you have to write the ON conditions correctly, matching not just C, but six aliased columns correctly three times. Ugh.

A much better solution is to use something like the CROSSTAB feature that's built into Microsoft Access. The query is dead simple and the software takes care of arranging the results across the page.

Alternatively, you could write a stored procedure. The procedure would run a simple query like

select C, PRICE
   from V 
order by C

The stored procedure would then use a cursor to loop over the result set, and use "current/previous" logic to detect when the current row has the same value of C as the previous row. The PRICE values would be written into adjacent columns of a temporary table during the looping. Finally, the procedure would return the contents of the temporary table.

For More Information

Dig Deeper on Oracle and SQL

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.