Hi, Disclaimer: I am a PostgreSQL user and consider myself a SQLite newbie.
I have a program that does mostly? SELECT requests, and it is very slow. But I then figured out that, if I rebuild my SQLite database without PRIMARY KEY/UNIQUE constraints, the program runs much faster (no measurement yet, but I?d say at least 10? faster). As I understand it, SQLite builds implicit indexes for PRIMARY KEY/UNIQUE constraints, but I would not expect those indexes to significantly decrease the performance of SELECT requests? For what it?s worth, my program is written in Python 3, and run on an up-to-date Debian Wheezy system, with Python 3.2.3-7, dynamically linked with libsqlite3 3.7.13-1+deb7u1. Is there anything well-known that explains this performance difference? Thanks for your help, -- Nicolas Boullis Footnote: ? The program first creates and fills 2 temporary tables, with locally-gathered data, and then only performs SELECT queries with both the 2 temporary tables and the permanent tables. Nothing is ever written to the permanent tables, and the temporary tables are created with no contraint.