High/low value identify

techsavy · Post by **techsavy** » Thu May 07, 2015 7:25 am

If (Convert(CHAR(255),'', Trim(A))='')
OR (Convert(CHAR(0),'', Trim(A))='')
Then 0 Else 1

Col A reading in CFF as binary

chulett · Post by **chulett** » Thu May 07, 2015 7:38 am

Convert only works on strings, as does your comparison... and trim.

techsavy · Post by **techsavy** » Thu May 07, 2015 8:02 am

okay got it. So is there anyway to do with a binary column? column is a COMP field as per copybook.

techsavy · Post by **techsavy** » Sun May 10, 2015 9:50 pm

Any clue friends?
I want to detect high/low field in a binary field in datastage.

ray.wurlod · Post by **ray.wurlod** » Sun May 10, 2015 11:57 pm

Use a RawToString() conversion.

chulett · Post by **chulett** » Mon May 11, 2015 7:49 am

That or I would think you could build equivalent binary stage variables to check against.

techsavy · Post by **techsavy** » Mon May 11, 2015 11:05 am

Craig- binary stage variable means?

chulett · Post by **chulett** » Mon May 11, 2015 12:08 pm

Basically it means a stage variable of the same data type and size as the incoming values you want to compare it to. And then you define one to have an initial value of all low values and another to have all high values as essentially constants during the job run. Then you would compare your incoming values to the stage variables. This rather than converting everything to strings.

In pondering this I'm not sure it is actually possible as I don't recall what data types are supported for stage variables. But at a high level, as generic advice, I've done things like that many times.

techsavy · Post by **techsavy** » Sun May 17, 2015 3:44 am

Actually I tried with RawTostring() but no luck. I am not sure when I am doing a view on source file in CFF the high value looks like '-1'.

Mike · Post by **Mike** » Sun May 17, 2015 8:56 am

It seems like the CFF stage has done all of the work for you. Sounds to me like your binary input has been converted to a signed integer. In that case, low values (all bits off) will have an integer value of 0, and high values (all bits on) will have an integer value of -1.

Mike

techsavy · Post by **techsavy** » Sun May 17, 2015 11:54 am

Looks like that Mike. but I am just wondering what is the decimal value to use for high value in job.

FranklinE · Post by **FranklinE** » Mon May 18, 2015 8:07 am

Now that I know this is a COMP field, the situation is different.

COMP is a storage format for a numeric value. It is not convertible to string in the usual sense, nor do you really want to think of it as "binary" data except in the FTP protocol sense.

You have a field of a certain length. Signed or unsigned, it has a minimum value of 0 and a maximum value of dec15 (F) in each position. Your check should be on that maximum value, not on the hex codes.

PIC 9(4) COMP has a maximum value of 9,999. One approach would be a logic statement like

Code: Select all

If Abs(col1) > 0 And Abs(col1) < 9999

would give you an active filter for all values in the column that you want to pass on, and a reject link could take the rest.

The odd thing about this is finding all xF in a COMP field. This is not standard practice that I'm aware of. I suggest checking back with your data source folks to find out why this is being done. In the meantime, to bypass the confusion, look to the actual storage of the field, and work your logic from that. If the COMP value is stored in a 2-byte storage length, then your maximum value is 65,535 (xFFFF).

techsavy · Post by **techsavy** » Tue May 19, 2015 6:15 am

Thanks Franklin. This is really usefull info. So shall I need to find out min and max for each such field?
As you said its a COMP filed, import as INTEGER in CFF stage.

FranklinE · Post by **FranklinE** » Thu May 21, 2015 9:15 am

The short answer is yes, that would be my choice. In the meantime, I believe you have a design issue with the high-values being set in a COMP field. I suggest you follow up on that with your Cobol team.