> > Comments? > > In one of your previous emails you've mentioned getting rid of prefixes > like 'The' and 'A'. I understand this is a twisted issue as you'd need a > list of all possible prefixes in all languages, important nevertheless.
That's a completely fair point. The title-collation is somehow in the air right now, but it's true that if implemented we need to consider not only the English case. > IIUC this is only possible in approach 3) and IMO needs to be taken into > consideration. > Not really. In all the cases, we could consider several collation functions, the default one as given by the Unicode support libraries, and an extra one implementing title collation. In approach 1) we could then have "COLLATE UNICODE" and "COLLATE TITLE" for example, and in each case we would use the corresponding collation function. For approach 2) we could implement the default Unicode collation for the indexes, but enable the use of additional collation functions as in approach 1), so merging both approaches: default collation would be Unicode collation and if needed, the COLLATE keyword could be added in the queries to specify another collation. In approach 3) we would also need to merge the approach 1), so we could have Unicode-based collation keys in the extra column and enable the COLLATE keyword in queries to allow other collations... -- Aleksander _______________________________________________ tracker-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/tracker-list
