I think Piers Angliss nailed it (once again): > If ANS is a very long string of numeric characters it will cause this. > It will pass the pattern match (and the other tests) because > each digit is numeric, but as a whole it is not considered a number
I tested it and that will cause the error. Furthermore, I have had a similar rash of errors lately where a few users must have asticky keyboard or something. Their ANS gets sometimes get tested as a file id, & errors out because > 256 character limit. So this one would be consistent with that. > Iirc, you had another of these around a year ago but I don't > recall the resolution Yes, it was similar and I believe you solved that one for me, too, Piers. I think it involved numeric strings with spaces in it passing the NUM( ) test, but failing when actually used as a number. I tried to look it up on the list archive, but the search is down. By the way, that bug was in a sister program to this one. _________________________ Let me respond to the other suggestions. Some overlapped & I think I cover averything here. And thank-you to those who answered. _________________________ > It looks like ANS is user input. yes > Any chance there are non-printable chars in there? yes < Although I wouldn't expect them to pass through either. > Maybe your UV version allows one of the above through??? I don't think so. _________________________ > We get this error when users enter "1.x" or "1X" or "2B", etc > on code like that. won't pass the MATCHES '0N' test. _________________________ > That would be my guess, that the line number isn't right. I *think* if source and object line up with VLIST, the line numbers are ok. Has anyone seen otherwise. _________________________ > any includes? 1. but I think it's clean. Nested includes and comments at the end of an INCLUDE line have been known to mess up line numbers. E.g.: INCLUDE PGMS ABC.H ;* this goofs up line counts But I don't know if that's still true at 10.0. _________________________ > what was entered when the message rendered? don't know. I caught this after the fact in uv/errlog. _________________________ > Only thing I can think of off-hand is that ANS = "+", > maybe "-", though I'm not sure if UV or various flavors > consider those values greater than numeric zero. Nope. I tested to be sure. They don't pass MATCHES '0N' test. _________________________ > fei: > Aix 5.3. Uv 10.1.12 > > ANS='0+' or ANS='0++' > > Both will match '0N'. And ANS is > 0. And you'll get the > non-numeric > for the extraction. YEOW! I tested this. and it's ok for me on 10.0.16, HPUX. Sounds like you've got a bug to report! _________________________ > On UD (5.1.27 and 6.0.12) char(0) and char(253) are valid "numbers". tested. not a problem here. ___________________________ Thanks again, everyone. Chuck Stevenson ------- u2-users mailing list u2-users@listserver.u2ug.org To unsubscribe please visit http://listserver.u2ug.org/