Originally posted by: SystemAdmin
I have a fairly simple partition type tree for a file that I want to read in bursts of 500 records. (File is <NL> delimited with three character initiator for each line. File has had rejects removed.)
When I turn burst mode off, the file is validated correctly. When I turn burst on, it fails with
i:835fd5194f"Input type contains errors."[/i:835fd5194f] warning message. The file has roughly 1800 records but in looking at the audit, I see that burst count went up to five. The trace shows that after the fourth
i:835fd5194f"End of Validation messages for INPUT CARD 1."[/i:835fd5194f] message there are a lot of trace messages, like this one:
code:1:835fd5194f(Level 2: Offset 3, len 0, comp 1 of 21, #1, DI 00000001:)
Data at offset 3 ('(end of data)') is INVALID data of TYPE
X'000F' (RIC_TO DZF_Columns Data).[/code:1:835fd5194f]
So it seems like the burst iterates one too many times. The "
b:835fd5194f(end of data)[/b:835fd5194f]" bit says the last time there is no data.
Actually I did some iterations of the map, tinkering with the fetch unit and found that it failed if the fetch unit was 460 to 532. Outside of that, it ran fine, though I didn't test much outside that range.
What is going on here?
I really need to output in
i:835fd5194fchunks[/i:835fd5194f] of 500 records. If someone knows how to get burst working, or a better way outside of burst, I certainly would enjoy hearing about it.
Thanks,
John
#DataExchange#IBMSterlingTransformationExtender#IBM-Websphere-Transformation-Extender