Actually, a script task doesn’t really work on a single record. It is simply that the example has you read in one record at a time.
For example, I had to create a special CSV script to replace the CSV script that snaplogic comes with. That script reads in, and keeps, the first row as a header, and reads the others using values from the first row. That first row could easily have been a thousand, and joining with the first row could have been processing all the data in the snap.
The size limit is based on your system, and various things like how soon you want it. Unfortunately, I think a lot of memory is not freed until you finish, though I could be wrong. Of course you should try to limit data requirements.
Basically, the snap has five processing points.
- Before you read any data.
- As you read the data.
- At some breakpoint you determine.
- Errors/exceptions as they occur.
- At the end.
And you can write out the records to any area as you want on your conditions and timing.
Steve