On 2017/03/02 2:29 PM, Deon Brewis wrote:
"This Query (for instance) will be exceedingly fast: SELECT ExtraCol FROM Foo WHERE UniqueCol > range_start AND UniqueCol < range_end" No, that's not covered. I've tried that before, that query is too slow when it isn't covered - the table is many GB's and the result is needed in low milliseconds. Also, I don't need an index directly on ExtraCol like specified below for the ( WHERE ExtraCol = ...) case . It's not useful a useful lookup by itself - only a useful result. So the only reason to include it in an index is to make it covered.
Ah yes, so you /do/ need every bit of work cycle saving you can get....
"You cannot optimize for everything, pick your favourite thing and optimize for that." This is a bit of a call to support INCLUDE columns in indexes. This optimization is very easy in SQL Server using that.
Well yes, so in SQLite you can optimize either for the speed or for the space, not for both. In MSSQL you can optimize for both using INCLUDE, and I can see how it is useful in specifically your case.
That said, I doubt it would make it into development code for SQLite at this point for the simple reason that it adds weight to the engine (adding INCLUDE) for a very fringe-case optimization.
Perhaps there are other reasons to add INCLUDE that are less fringe-case? _______________________________________________ sqlite-users mailing list [email protected] http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
