Thanks to Wolfgang for spotting the problem. It was not the inlist iterator at all but
an outer join! The NOT NULL predicate invalidated the outer join, so the optimizer was
smart enough to make a different decision. I am still perplexed as to why the table
access information was so radically different, but in light of the new finding, an
explanation can be found.
The lesson learned here is to not to focus on a 'problem' until you fully understand
the whole situation.
Daniel
begin:vcard
n:Fink;Daniel
tel;cell:303.808.3282
tel;work:303.272.3225
x-mozilla-html:TRUE
adr:;;;;;;
version:2.1
email;internet:[EMAIL PROTECTED]
title:DB Services Lead
x-mozilla-cpt:;-4832
fn:Daniel Fink
end:vcard