Now *here's* an odd problem with dclerk

[Bruce Easton]

Jay R. Ashworth wrote Monday, March 17, 2008 6:18 PM:
>
> On one machine that I and a contractor are working with, dclerk 5.0.13
> displays the following problem *on some databases and not others*
> (specifically, ones he created recently seem to do it, and older ones
> do not -- the new ones *may* have been ported in from Windows by xfer,
> and I'd need to double check the version on his laptop):
>
> Record NNN has 'N' in field 44, confirmed using dscreen and dclerk.
> (We had just posted the data there, but it never hurts to check)
>
> Build an index on 44, search for 'N', confirm that record NNN is
> displayed (in the browse).
>
> Use short or extended selection to search for records where 44 eq 'N'.
> No records are returned at all.
>
> Just for grins, I turned index scan off with -jn, but that didn't help.
>
> Anyone ever seen this?
>
> Cheers,
> -- jra
> --
> Jay R. Ashworth                   Baylink
>

Jay - not sure what you mean by confirmed using dscreen and dclerk,
specifically the reference to dscreen.   Maybe you are saying you've
confirmed with a real field indicator on a screen and not just the
browse that the field is populated with 'N'?

I hate to get nit-pickety, but ... the field that holds 'N' is
one byte, right?

And yes, thinking about Jeff's question, depending on the unexpected
behavior you are experiencing with browse vs. screen (if that's what appears
odd), an auto table could be involved.  Easy enough to test -
running without the auto table.

Bruce

Bruce Easton
STN, Inc.