[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: ordered indexing for integers



Russ Allbery wrote:
I say go for it.  It's early in the 2.4 lifecycle, not many people will
have upgraded yet, and I expect a lot of people, when upgrading to 2.4, to
take advantage of the opportunity to go to a newer version of BerkeleyDB
and therefore have to dump and restore their database anyway.
(Distributions in particular will want to try to minimize the number of
BerkeleyDB libraries they keep around, and if 4.6 works properly, there's
a strong motivation to drop 4.2.)

Good enough. OK, it's now in HEAD. The feature is disabled by default (so backward compatibility is still there by default); you have to configure index_intlen to enable it. The setting specifies how many bytes of an integer to use as the key value. The default is zero, which disables it. (Deleting the setting with ldapmodify also disables it.)

The minimum setting is 4, which will use the first 4 bytes of an integer. The largest value would be 255; presumably no one will ever have integers that long but whatever...
--
-- Howard Chu
Chief Architect, Symas Corp. http://www.symas.com
Director, Highland Sun http://highlandsun.com/hyc/
Chief Architect, OpenLDAP http://www.openldap.org/project/