Cetin
We have a state controled mainframe that houses all our accounting data that can be batch updated through overnight processing. The structure we use is a fixed 52 field table that is converted to ASC at the mainframe. That was not my problem. There 2 tables, one for transaction and one for corrections. There are 3 more fields in the corrections table in different locations.

I solved the problem with a case statment that tested transacation table for the field type and size. if that is correct, I creat the corrections table from the transaction table with an SQL query creating space for the fields in the correct place. The new fields do not have data, they just need to be there for the structure, otherwise it blows at the mainframe.

Anyway, problem solved, thanks for the interest.

>>We have to have a table with fields in a specific order to upload to a mainframe. The field names do not matter, just the location of the fields. I want to be able to take any table that has 52 fields with each having the correct data type and size and add 4 fields to it in a specific place. I looked a the FAQ and one answer I found referred to a bubble sort but that is to complicated for the number of fields. The problem is other table fields may have different names and record size is too large for an array (which is what got me here in the first place). I currently use SQl with space() as fieldname to add fields, into array and append from array, but data is too large for array. Any ideas?
>>
>>Thanks, Chuck
>
>Chuck,
>"fields in a specific order to upload to a mainframe. The field names do not matter, just the location of the fields"
>"record size is too large for an array"
>
>I assume performance would then matter too, right ? Saying mainframe what would you send and how ? SDF, Delimited or a free table ?
>Would you elaborate this more. I have a great feeling the answer is hidden in 'set fields' :)
>Cetin