Several of the
DATA step functions use two-dimensional arrays for input or output.
The arrays enable applications to move information in and out of
the metadata server with fewer calls. However, the DATA step is not
two-dimensional, so the following conventions enable you to handle
these multiple-row arrays:
-
For functions that return arrays,
the function asks the metadata server to return only one row (or a
specific row) of an output array. The output array is generally kept
in an object cache that lasts only as long as the DATA step. The
key to the cache is the
uri argument, and the key to the row is the
n argument. When you submit the function, it checks whether information
from the output array already exists in the cache and, if so, returns
the information from the cache. If the information does not exist
in the cache, the function calls the metadata server to fill the cache.
You can use the
n argument
to iterate through the rows of the array; see how
n is used in
Examples: DATA Step Functions for Reading Metadata and
Examples: DATA Step Functions for Metadata Security Administration.
-
The functions that input arrays
are similar to the functions that return arrays, but the array is
not kept in an object cache. Rather than iterating with an
n argument, you specify the multiple values in
a comma-delimited list. In some functions, you submit two values that
must be in parallel. In other words, for a
name,
value pair, if you specify
three
name arguments, then
you must specify three
value arguments.
For more information
about DO loops and array processing in a DATA step, see
SAS Language Reference: Concepts.