Re: [Bug-GnuCOBOL] Incorrect parsing of numeric-edited values

[Ron Norman]

I think that test case is perfectly correct and no warning is needed.

I would expect that ZERO ZEROES ZEROS ZEROE 0 0.000 are all meaning a value of 'zero' so

shouldn't the following all be accepted and mean that the initial value of the field is zero.

01 N PIC .999 VALUE 0.

01 N PIC .999 VALUE ZERO.

FYI. Micro Focus does not give any warning with this.

And if it is VALUE ZERO then GnuCOBOL also gives no warning.

I think 0 is the exact same as ZERO.

On Sat, May 13, 2017 at 7:19 AM, Sergey Kashyrin <address@hidden> wrote:

Hi Edward,

I see GC is accepting the value. Just giving a warning which is reasonable.
Don't think we need to eliminate the warning, which will happen with this patch.
Maybe we need to proceed as with numeric, but to give a warning, at least with std=ibm or osvs

SK

On 5/13/17 6:55 AM, Edward Hart wrote:

Hi David,

Thanks for the bug report and patch. This is technically extension, but it's a very innocuous one and it's supported by Micro Focus (see VALUE Clause, General Rule 1.b). I'll commit this immediately.

Edward

On 13 May 2017 at 10:56, David Newall <address@hidden> wrote:

Hi all,

I believe it's bug:

address@hidden cat test-pic.cbl
        IDENTIFICATION DIVISION.
        PROGRAM-ID. TEST-PIC.

        DATA DIVISION.
        WORKING-STORAGE SECTION.
        01  N    PIC .999 VALUE 0.

        PROCEDURE DIVISION.
               START-PROCEDURE SECTION.
        000-TOP.
            DISPLAY 'N:', N.

        END PROGRAM TEST-PIC.

address@hidden cobc -x test-pic.cbl
test-pic.cbl: 6: warning: alphanumeric value is expected

If I'm right (about it being a bug), the problem is in typck.c; and because of the "TODO" comment, I think it's something that slipped through the cracks.  I think NUMERIC-EDITED could be parsed the same as NUMERIC is.

I'm using r1560.  I've attached my suggested patch.

Cheers,

David

--

Cheers

Ron Norman