I want to know how functional indexes are used "in the real world". Here
are the common uses:
* non-unique index on the first parts of a longish text field
(SUBSTRING(field)) to save disk space, while still allowing faster
searches than a sequential scan.
* indexing on LOWER(field)/UPPER(field) to allow case-insensitive
searches or case-insensitive unique constraint.
In what other scenario are you using functional index? Searching the
archives, I found someone indexing a date as 'YYYYMMDD', probably to
allow searching using LIKE 'YYYY%' or LIKE 'YYYYMM%'. I've also thought:
* creating a REVERSE_DOT(field) function for IP addresses and hostnames
to allow wildcard searches (*.yahoo.com) using 'REVERSE_DOT(f) LIKE
'com.yahoo.%'.
* creating a HOSTPART(field) and LOCALPART(field) for email addressess;
Has anyone done something like this or other tricks using functional
indexes? It seems potential to simplify data model (no need to store
email addresses as host + part, etc).
--
dave
