IS NUMERIC check for SPACES
in [Cobol]
|
Prev: My COBDATA problem
Next: License Management Service
From: Anonymous on 3 Dec 2009 15:16 In article <hf94sb$aql$2(a)news.eternal-september.org>, Doug Miller <spambait(a)milmac.com> wrote: >In article <hf5t5t$og3$1(a)reader1.panix.com>, docdwarf(a)panix.com () wrote: >>Now my memory is, admittedly, porous but I recall that IBM mainframe >>compilers (IKFCBL00 and IGYCRCTL) will successfully test COMP-3 (packed >>decimal format) fields with IS NUMERIC... and as a COMP-3 field, in my >>experience, often has its least significant nibble reserved for the >>field's sign (usually C for positive, D for negative) then a COMP-3 field >>containing X'123D' may pass an IS NUMERIC test while not, by your >>standard, containing an numeric digit. > >1, 2, and 3 are numeric digits, no? The sign nybble is ignored. If the sign is ignored then a COMP-3 field containing X'0000' would pass an IS NUMERIC test... and as I recall - porous memory and all - it doesn't. COMP (binary) fields, no problem... but COMP-3 needs a valid sign. Perhaps some testing is in order. DD
From: slade on 11 Dec 2009 22:47 On Dec 3, 3:16 pm, docdw...(a)panix.com () wrote: > In article <hf94sb$aq...(a)news.eternal-september.org>, > > Doug Miller <spamb...(a)milmac.com> wrote: > >In article <hf5t5t$og...(a)reader1.panix.com>, docdw...(a)panix.com () wrote: > >>Now my memory is, admittedly, porous but I recall that IBM mainframe > >>compilers (IKFCBL00 and IGYCRCTL) will successfully test COMP-3 (packed > >>decimal format) fields with IS NUMERIC... and as a COMP-3 field, in my > >>experience, often has its least significant nibble reserved for the > >>field's sign (usually C for positive, D for negative) then a COMP-3 field > >>containing X'123D' may pass an IS NUMERIC test while not, by your > >>standard, containing an numeric digit. > > >1, 2, and 3 are numeric digits, no? The sign nybble is ignored. > > If the sign is ignored then a COMP-3 field containing X'0000' would pass > an IS NUMERIC test... and as I recall - porous memory and all - it > doesn't. COMP (binary) fields, no problem... but COMP-3 needs a valid > sign. > > Perhaps some testing is in order. > > DD In an IBM mainframe environment, depending on the NUMPROC/NUMCLS COBOL compiler options selected, your sign choices are limited, at best, to X'A" thru X'F'; at worst, to X'F'. Any other nybble will result in a "FALSE" return from IF NUMERIC. So, as Doc said, the sign for numeric DISPLAY (and COMP-3) fields is not ignored; X'F1F203' will generate a "FALSE" return. If the item is defined w/SIGN IS SEPARATE (not an issue here, apparently) there are other criteria.
From: Anonymous on 12 Dec 2009 11:00
In article <2a82e1bc-8e40-4704-8206-c76ee94a5f0a(a)n13g2000vbe.googlegroups.com>, slade <jnjsle1(a)optonline.net> wrote: >On Dec 3, 3:16?pm, docdw...(a)panix.com () wrote: >> In article <hf94sb$aq...(a)news.eternal-september.org>, >> >> Doug Miller <spamb...(a)milmac.com> wrote: >> >In article <hf5t5t$og...(a)reader1.panix.com>, docdw...(a)panix.com () wrote: >> >>Now my memory is, admittedly, porous but I recall that IBM mainframe >> >>compilers (IKFCBL00 and IGYCRCTL) will successfully test COMP-3 (packed >> >>decimal format) fields with IS NUMERIC... and as a COMP-3 field, in my >> >>experience, often has its least significant nibble reserved for the >> >>field's sign (usually C for positive, D for negative) then a COMP-3 field >> >>containing X'123D' may pass an IS NUMERIC test while not, by your >> >>standard, containing an numeric digit. >> >> >1, 2, and 3 are numeric digits, no? The sign nybble is ignored. >> >> If the sign is ignored then a COMP-3 field containing X'0000' would pass >> an IS NUMERIC test... and as I recall - porous memory and all - it >> doesn't. ?COMP (binary) fields, no problem... but COMP-3 needs a valid >> sign. >> >> Perhaps some testing is in order. > >In an IBM mainframe environment, depending on the NUMPROC/NUMCLS COBOL >compiler options selected, your sign choices are limited, at best, to >X'A" thru X'F'; at worst, to X'F'. Any other nybble will result in a >"FALSE" return from IF NUMERIC. > >So, as Doc said, the sign for numeric DISPLAY (and COMP-3) fields is >not ignored; X'F1F203' will generate a "FALSE" return. That's as I recall it, aye... but it was from Long Ago and perhaps things had changed. Has anyone done any testing? If so, might they post the code and results, so that experiment's reproducibility be verified? DD |