lack of ordering support

[Jon Roberts <jon@mentata.com>]

In a recent design, I inadvertantly relied on the ability to perform >= 
and <= searches on an attribute that has no ordering. I searched the 
mailing list archives and found mention of this, then checked the schema 
and admitted defeat. This was my fault and I accept that, however...

Why is OpenLDAP so stingy with attribute ordering? In the core.schema 
I'm looking at (admittedly an old one) I see only 3 attributes with 
ordering defined, and two are timestamps. From what I read, simple 
numerical ordering isn't supported at all in distributions, even though 
it's listed in the documentation as if you could at least use it with 
custom attributes.

The only theories I have, since I can't find other mention on this as a 
deficiency, is that either the overhead is really high (I can't imagine 
why), or the available routines are really poor. If it's just a matter 
of sticking to the letter of the RFCs, I can't agree that this is necessary.

Jon Roberts
www.mentata.com