Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
I opened a DFSORT Application Programming Guide (I used 2.1) PDF and used the PDF-reader to search for 'mul,' (quotes not needed for the search) and readily located several examples of the use of multiple arithmetic operands.
Thanks Bill. I was expecting this reply.
I have 2012 version of DFSORT Application Programming Guide, I can't see any example for multiplication there.
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
If you don't specify a length (with LENGTH or an explicit EDIT mask) you get a default length. If you don't provide an explicit EDIT, you get a default mask.
"I" in an EDIT means "a blank if there is no significant digit encountered yet", so leading zeros you ask to be blank.
You need a + (or -) because that is how DFSORT will tell if you are specifying a numeric literal, or the (incomplete) start of a field definition.
I'm pretty sure the examples for 2.1 are the same as for 2.2, for instance, p136 in both documents:
Quote:
arexp,edit or (arexp),edit
specifies that the edited result of an arithmetic expression is to appear in the reformatted input record. The arithmetic expression can consist of input fields, decimal constants, operators and parentheses. An arithmetic expression produces a signed, 31-digit zoned decimal (ZD) result to be edited as specified.
See arexp,edit under OUTFIL OUTREC for details.
Joined: 17 Oct 2006 Posts: 2481 Location: @my desk
zh_lad,
I would suggest keeping all the intermediate multiplication results in 2 bytes, so you don't have to figure out later which result has 2 digits and which one has 1 digit. Because you need to sum all of them later. It might add more work and confusion if you get rid of the blanks, and then try to figure out where each intermediate result starts.
So you can use TT...edits and read them as ZD in further computations or keep the IT...edits and read them as UFF, whichever works for you.
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
And multiple arithmetic operations can be carried out without having to define anything for intermediate results. The next operation is on "the current result" for as long as necessary without having to be concerned about it.
1. I learnt it later when I read about Mask.
2. I am aware of use of 'I' in EDIT
3. I didn't know, I was using arithmetic first the first in SORT - Thanks
4. Yes PDF has examples but always problem is discussed on different page. Ideally it should have link to go directly to the problem description by one click. It difficult to absorb example (sample) if problem is discussed on a different page.
Thanks for your comments. They helped to use PDF better.
Bill Woodger wrote:
If you don't specify a length (with LENGTH or an explicit EDIT mask) you get a default length. If you don't provide an explicit EDIT, you get a default mask.
"I" in an EDIT means "a blank if there is no significant digit encountered yet", so leading zeros you ask to be blank.
You need a + (or -) because that is how DFSORT will tell if you are specifying a numeric literal, or the (incomplete) start of a field definition.
I'm pretty sure the examples for 2.1 are the same as for 2.2, for instance, p136 in both documents:
Quote:
arexp,edit or (arexp),edit
specifies that the edited result of an arithmetic expression is to appear in the reformatted input record. The arithmetic expression can consist of input fields, decimal constants, operators and parentheses. An arithmetic expression produces a signed, 31-digit zoned decimal (ZD) result to be edited as specified.
See arexp,edit under OUTFIL OUTREC for details.
Joined: 17 Oct 2006 Posts: 2481 Location: @my desk
Quote:
Thanks Rohit, it worked
But I am afraid that is not what you were trying to achieve. Like Bill mentioned above, you can have multiple computations, just use braces to have meaningful results and then edit the final result as needed.
You don't need to worry about the format of intermediate results because, as suggested you can do multiple computations one after the other, like this and edit only the final result.
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
IFTHEN=(WHEN=INIT for an OVERLAY to give you the several multiply-by-two results (they are not "intermediate results" in the sense normally used), then another WHEN=INIT with the BUILD to ADD all the digits that you need (odd-number from original location, even-numbered from the location OVERLAYed to) and then multiply that by +9.
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
No-one locked anything, and that last comment wasn't mine :-)
Posts can only be edited by non-moderators for 10 minutes after they have been entered. If you can post (or PM) me a correction I can apply it and tidy up a bit.
How will I deal with space when result of multiplication is only 1 digit.
Code:
1,1,UFF,MUL,+2,EDIT(IT)
How will I ignore space to pick up digits to sum in OUTREC (second IFTHEN).
Rohit was using FIND and REPLACE
Code:
OUTREC FINDREP=(INOUT=(C' ',C''))
Thanks.
Bill Woodger wrote:
IFTHEN=(WHEN=INIT for an OVERLAY to give you the several multiply-by-two results (they are not "intermediate results" in the sense normally used), then another WHEN=INIT with the BUILD to ADD all the digits that you need (odd-number from original location, even-numbered from the location OVERLAYed to) and then multiply that by +9.
Joined: 17 Oct 2006 Posts: 2481 Location: @my desk
I think you got confused. Forget the FINDREP suggestion. You don't need FINDREP. I will repeat what was suggested already if that makes any clearer.
1 - In the first IFTHEN , multiply only the digits at odd positions(pos-1,3,5,..etc), which in the input are in ZD format (you don't need to multiply others by 1) and OVERLAY at the end of the record (say pos-81,83,85..etc), here either keep it edited as 'TT', or 'IT' so you can read them as ZD or UFF in the second IFTHEN.
To avoid confusion I'd recommend using TT, so you see ZEROes instead of blanks.
2 - In the second IFTHEN, ADD all the individual digits from the results above(pos-81,82,83,...etc) and the digits at even positions from your original input which is still available at pos-2,4,6,..etc. And include the MUL by 9 in the same calculation, edit the final result to format it as per your requirement.
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
zh_lad wrote:
I was amazed to know every IFTHEN picks up amendments done by previous IFTHEN
It is not the IFTHEN, it is the BUILD. Each BUILD that is executed creates a "new version" of the current record, from the previous version.
OVERLAY does not create a new version of the current records, but operates on the existing version.
Code:
BUILD=(21,20,1,20)
OVERLAY=(21,20,1,20)
Try those, the output will be different, because of the way BUILD and OVERLAY operate.
If you need to "rearrange" data, BUILD is more convenient. If you need to update data where it is, OVERLAY is 1) clearer and 2) performs better.
Other differences are that although you can specify columns in BUILD, you can only define fields in "ascending" order (the columns will just get space-padded if needed), whereas with OVERLAY you can specify the columns in any order, and even do multiple transformations to the same field in one OVERLAY, by specifying the same column multiple times.
Anyway, I'm assuming that was the confusion (how BUILD works), because it is entirely normal to expect something that has been changed to be available to the next line of code.
Note also that with IFTHEN=(WHEN=(logicalexpression), when one is true, no further IFTHEN=(WHEN=(logicalexpression) are processed for that "record", unless you ask for it, by using HIT=NEXT, and that includes if you want to use WHEN=ANY.
So one or more IFTHEN=(WHEN=(logicalexpression) are like a "case/select" statement (EVALUATE in COBOL).