Given the following definitions:

    col1 int,
    col2 int,
    col3 int,
    col4 int

    @Param1 int,
    @Param2 int
    SELECT @Param1 * 2 AS 'res1', @Param2 * 4 AS 'res2', @Param1 AS 'col3'

I am trying to join the function to the table so that I get a calculated value (s) per row. Example: In a bookings table, I need to get the price based on a start date and end date.

Here is an example query:

SELECT      tbl.col1
,           tbl.col2
,           tbl.col3
,           fn.res1
,           fn.res1
FROM        MyTable tbl
CROSS APPLY fn_MyFunction(tbl.col3, tbl.col4) fn
WHERE       fn.col3 = tbl.col3

While I think I am getting the correct results in my actual query, how does the CROSS APPLY get the results in this case? Does it get them row-by-row (effectively)? Or does it work in a similar way to CROSS JOIN (ie: making a Cartesian Product)?


CROSS APPLY takes a table valued function and 'applies' parameters from each row in the query you are applying it to. The function is evaluated once for each row and the output is implicitly joined to the source row in the record set from which the parameters were obtained. Note that this 'join' can be 1:M - with a TVF one row at source can generate multiple rows of output from the TVF.

You don't need to use any more predicates in the where clause to join the results unless (for some reason) you want to further filter the output of the function by applying an additional predicate.

CROSS APPLY is quite useful with xpath queries if you have to go ferreting around XML fields that may contain repeating groups.

  • +1 - It's also useful for calculated fields that you want to reference more than once - you can use the alias instead of the whole definition.
    – JNK
    Apr 4 '12 at 14:46
  • @JNK - long time no see in the heap. Apr 4 '12 at 15:07
  • Yeah I got busy, (un)fortunately...
    – JNK
    Apr 4 '12 at 15:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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