bug: possibly related to bug previously submitted?

[John Esak]

Hi,
We (tony Ryder in Australia) and I found another bug .  He presented a
situation that dealt with a browse lookup and the pkeep function.  With a
standard browse lookup, an exit key of "D" that deleted the browse
records... everything would work properly until the very last record in the
file was deleted. In this case, the browse would not clear the last record
deleted. It would actually get deleted, but the record remained on the
screen as if it weren't.  I exactly duplicated his processing and it worked
on my system.

By the way, this is repeatable all the way up to 5.0.13, SCO Unix and
Windows.

Okay, we started doing things on his.... the most important, we thought, was
to change the pkeep to keep. This fixed the problem, the record deleted
_and_ was cleared from the browse.... we were left with the error "No more
records for this browse lookup".  the right behavior, but no pkeep
functionality.   We changed his variable definition  for the browse
parameters to a real "literal" browse .  No change. We mucked around with
the value of k. (And this is where I thought this bug might be related to
the one I submitted a couple months ago. If a k value is set to null, the
pkeep will _not_ work, the highlight always returns to the first line.
Setting the k value to the lowest possible index value makes it work.)   In
any case, this did not do anything to help.  So we spent another hour or two
playing around with values and constructions to no avail.

Finally, I looked at the code I was using which _exactly_ matched his and
well, the word exactly was not quite right.  My code had one simple
difference.  I had not specified a length for the k value. He had.  I had
not noticed this because his variable definition was much further up in his
code.  Anyway, the difference of doing...


if:
then: ky=""

and

if:
then: ky(8)=""

made the difference!  With the variable given a defined length, the last
record in a file when deleted with a browse key, does actually disappear
from the window.  Now, here is the rub.

The bug I submitted earlier says that if the k value is "" then the browse
will not do the pkeep... SO, you are caught between a rock and a hard place.
Do you want the pkeep to function properly, or do you want the last deleted
record in a file to disappear correctly from the window?  Take your choice,
because you get either or.

Therefore, I'm submitting this bug as an addendum to the previous one in
hopes that the whole relationship between the k value and the pkeep
function (and now the deleted/displayed/last-record thing) and any other
consideration all be dealt with comprehensively. Obviously, the
functionality should not be exclusive.

Should anyone at FP Tech need more information on this, please contact me...
or come to the FP Room and we can show you the problem in great detail.

Thanks,

John Esak

Visit The FP Room at www.tinyurl.com/yuag7 24/7