Kurt D. Zeilenga wrote:
It might be worthwhile to use present indices where
no keys substrings keys are found. Could also use
it narrow ordering to just those entries where the
attribute is present.
Such a change might have messed up things for us, if we needed presence
indices.
We will be using sizelimit size.unchecked to make too general queries
fail at once:
- partly so users won't have to sit wait for a while and then just get
timelimitExceeded anyway,
- partly for privacy protection,
- partly for the sake of server speed.
We have found that we'll need to set the unchecked limit uncomfortably
high for the sake of substring searches, but at least it does still cut
off a lot of such searches. Possibly we'll add a bunch of invisible
dummy entries to make it fail more reliably, or something. Anyway, that
high limit means that with the above change, a lot more queries will
pass the unchecked test only to fail the filter.
Howard Chu writes:
Of course, in practice, I never use
presence indices. I wonder how many people do.
Probably quite a lot, since I think some OpenLDAP documentation says or
said that if one has an equality index (I think), one might just as well
add a presence index, since that takes very little extra space. I
believe that's why we had presence indices.