> 2 dec. 2018, E.Pasma: > >> 30 nov. 2018, AJ Miles: >> >> Ah, this tool seems very handy. For those curious, I'll paste the results >> below. The index approximately doubles the storage size, but I am >> intentionally making that tradeoff to avoid the slow down when enforcing a >> unique/primary key on the Reference table while inserting. >> >> -AJ > > Hello, > > is the "integer" column unique and a potential primary key? Then it would be > surprising that defining it as such causes a slowdown. Assuming the table is > then defined WITHOUT ROWID. And that the multithreading pragma is set. > > About sqlite_analyzer: on macOS 10.13.6 I got > ..Tcl.framework/Versions/8.4.. image not found > After editing the binary file as below, it is alright. > vi sqlite3_analyzer > :%s/8\.4/8.5/ > :x > > Thanks for the intriguing subject. E. Pasma. Sorry,
In a small test, I measured inserting in a table with an INTEGER PRIMARY KEY and without rowid. Unfortunately, it causes a slowdown of a factor 5, compared to your schema. In my test, the rows are inserted in random order. If inserted in the order of the key value, the result is different altogether. But it looks hard to achieve that with billions of rows. Thanks, E. Pasma _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
