Some interesting behavior was discovered while attempting simultaneous queries to a MS SQL server through the ODBC/freeTDS layer on SUSE on a persistent connection with ODBC cursors enabled. Stepping through the first resource with odbc_fetch_array() in a while loop works as expected, but calls to subsequently created resources within that loop were failing to return data where MySQL equivalents to the script were working just fine.
I found that adding a num_rows function call [custom function that simply steps through the result set... the misbehavior of odbc_num_rows() is another matter] immediately following each odbc_exec() then resetting the cursor seemed to cache the results for proper performance of the multiple active result sets. It may be slightly slower than the MySQL equivalent, but at least it works.
