Re: READE implementation difficulties. RE: When is %EOF not an %EOF

["Buck" <buck.calabro@xxxxxxxxxxxx>]

> When I said "compiled program code" I mean whatever
> the compiler does to implement READE.  It would read
> the next sequential record, compare the keys
> and then load the rcdfmt fields that the user code sees.

And the light comes on.  So you're talking about an internal mirror (using
Ext DS?) of the database fields which only get populated on our good match,
not our pseudo EOF.  Interesting.  I don't see why that wouldn't work, and
considering that this is a performance thread, it'd be interesting to find
out how our ersatz READE performance compares to the compiler's
implementation of READE vs. READ without the duplicate DB fields.  Oh, the
cool questions that the boss won't pay to have answered...  Another one for
after hours.
  --buck