Updating postgressql for hold em manager
This is especially important now that people are active developing Oracle and MS SQL abstraction layers for Drupal.
An extra column, on the other hand, is something that can be used across the board.we're talking 4 or 5 orders of magnitude faster, here. i agree w/ killes, this doesn't strike me as particularly hacky at all, and storing an already-LOWER() copy of name seems much more portable than playing with various DB-specific hacks. personally, i think storing the LOWER(name) as a seperate column is just fine, and will definitely be the easiest to port (if slightly inefficient w/ space in the DB). kind of like $connection, but with the $db_type (or do we have this?
And the goal of creating the index that way would have to be to force Postgres into behaving like My SQL (ie case insensitive search).
It seems like the best solution would be if Postgres had some setting that would configure the behavior of their string comparison operator.
And only a matter of time before someone does an SQLite port, or a Sybase port, or a DB2, etc.
I'm not against a db specific approach, especially if we can apply it to the other major systems I mentioned.
But the "hackishness" or "tackiness" or "this hurts my normalized eyes-ness" of the patch shouldn't prevent us from harvesting the performance gain.
Also, if this feature needs to be retained, it might be more elegant to use Pg SQL's operator overloading feature to accomplish this.
Not sure if it's possible, but I'll try and ask Cvbge about it. Cheers, -KIt would be good to research some of the other rdbms solutions out there as well and see what issues we might run into if we do an Oracle or MSSQL port.Forming the index differently won't correct the problem, only taking the function off of the column name can help.