ISPF tables can only be stored in PDS members. But their structure does not fit into the record structure of this file system. The number of fields might not be fixed with all table records. In addition the length of the fields is never fixed.
This causes ISPF to store table data in a "stream-like" format with some header data followed by the record data with each column separated by special characters (x'00') and a hex length field introducing and identifying the following data as column contents. The data of each table row is written up to the length of the current record and then continued in the next record.
So it is impossible to use a simple file compare utility which assumes that every record has the same layout.
I recommend exporting the table data to two datasets with the ISPF Table Utility (usually option 3.16) and then run a compare on the resulting datasets.
If you need to do it programmatically I'm pretty sure that you have to follow TheMFKid's advice.
Joined: 01 Sep 2006 Posts: 2029 Location: Silicon Valley
compare the values of column1 in table A with values of column2 in table B
Stefan and theMFkid are right, but even then the col1 and col2 requirement makes the task a little harder. Even if there was a tool to compare table to table, you might still need to write your own to compare col to col.
Tables should be members in a PDS, so what's the problem
As Stefan pointed out, the table structure does not map well in to FB80 records. A change in the first row will cause a change in all subsequent records, making a normal member comparison meaningless.