On Sun, Apr 3, 2011 at 6:35 PM, Clint Byrum <[email protected]> wrote: > Excerpts from Brian Aker's message of Sat Apr 02 18:13:36 -0700 2011: >> Hi! >> >> For latin1 and swe7 should we accept them as character set specifiers for >> ease of use? I believe they are a subset of UTF-8. > > As a sub-concern.. utf-8 leads to 3-bytes-per-position indexes right > now. I have to wonder if it would be easy to create a new index type that > only indexes 2-byte chars for situations where that is acceptable. The > question of what to do w/ 3 byte chars would need some thought, but > I think my first inclination would be that they would be rejected, > or possibly just stripped out (meaning unique indexes and index scans > would no longer be useful).
IMO such an optimization should be an implementation detail not observable to the user. -- Olaf _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
