Since Powerbuilder 5 introduced datastores I have found myself using them instead of cursors.

Advantages

• Datastores are easy to use for anyone who has used a datawindow.
• Datastores are efficient at retrieving data.
• You can preview the data for debugging.
• The Powerscript code is slightly simpler
• You can do grouping easier.
• You can add advanced filters.
• You don't always need variables for your data.
• You can add computed fields to the datawindows and use the answers.
• You can find the rows count without a separate select.
• Usually faster (from my experience)

Disadvantages

• You need the datawindow object.
• You have to make sure you destroy your datastore to avoid memory problems.

Here is some sample code of a cursor

string ls_cust_code, ls_customer_name, ls_town

DECLARE lc_my_cursor CURSOR FOR
   SELECT customers.cust_code,
          customers.customer_name,
          customers.town
    FROM customers
    ORDER BY cust_code;

OPEN lc_my_cursor;

    // Fetch first row
FETCH lc_my_cursor INTO 
    :ls_cust_code,
    :ls_customer_name,
    :ls_town;

DO WHILE SQLCA.SQLCode = 0
    // Loop code
    // Fetch next row

   FETCH lc_my_cursor INTO
       :ls_cust_code,
       :ls_customer_name,
       :ls_town;
LOOP

CLOSE lc_my_cursor;

And the equivalent datastore code

string    ls_cust_code, ls_customer_name, ls_town
long      ll_row, ll_row_count
datastore lds_datastore

lds_datastore = CREATE datastore
lds_datastore.DataObject = "ds_customers"
lds_datastore.SetTransObject(SQLCA)
lds_datastore.Retrieve()

ll_row_count = lds_datastore.RowCount()

FOR ll_row = 1 TO ll_row_count
   ls_cust_code = lds_datastore.GetItemString(ll_row, "cust_code")
   ls_customer_name = lds_datastore.GetItemString(ll_row, "customer_name")
   ls_town = lds_datastore.GetItemString(ll_row, "town")

   // Loop code
NEXT

DESTROY lds_datastore