Reading timestamp in sequential file

qt_ky · Post by **qt_ky** » Wed Dec 10, 2014 9:55 pm

Rejecting can be a two step process.

Records in your example will get rejected when the timestamp values to not match the format tag you specify, such as when a day number or month number has too many or too few characters.

When the timestamp value matches the format tag, the record will be passed to the next stage, whether or not the timestamp is valid. If it is invalid, as in month 22 or day 33, it should get set to the default value, which is an optional property you can find and set when you double click the column.

Go with your latter format tag followed by a Transformer stage where you can do further tests, like IsValid() or IsValidTimestamp(), and reject records from there also.

abc123 · Post by **abc123** » Wed Jan 07, 2015 10:40 pm

Eric, that still doesn't answer his main question which is that the format string "%(m,s)/%(d,s)/%yyyy %hh:%nn:%ss" should cause the 2nd record to fail and it doesn't.

ray.wurlod · Post by **ray.wurlod** » Thu Jan 08, 2015 2:43 am

It appears that both (d,s) and (m,s) are checking only for a one- or two-digit number, without performing any range check.

If you think this is a bug, please contact IBM, get their take on it, and let us know.

Anyone else like to test it using other versions?

qt_ky · Post by **qt_ky** » Thu Jan 08, 2015 7:41 am

abc123: The 2nd format tag is more accepting than the first... Please read my answer a few more times, test it yourself, and compare the results you see with the documentation about the format tags.