#!/bin/sh

integrity_check() {
/usr/bin/sqlite3 “$HOME/Library/Mail/Envelope Index” “PRAGMA integrity_check;” 2>&1
}

case “`integrity_check`” in
ok) ;;
*)
echo “Integrity check failed.” >&2
echo “$integrity_check” >&2
exit 1
;;
esac

/usr/bin/sqlite3 “$HOME/Library/Mail/Envelope Index” “vacuum;”

[...][...]

None of the iTunes slowness you describe is due to any kind of database issue. As a matter of fact, iTunes’ database needs are really minor. It only has to search on a fairly limited and highly compressable set of metadata to perform the database oriented operations of the UI. iTunes database, by and large, can easily fit in memory without putting undue stress on the system.

Could iTunes use SQLite and gain a performance improvement? Maybe so, probably not though, certainly, it would gain scalability.

The kinds of “refine as user types” searches that iTunes uses really do not lend themselves to SQL based searches unless you do a lot of optimization work. If your data set is small enough to fit into memory (with some likely pruning of less commonly searched information), you are far better off optimizing your search-as-type around in memory data structure traversal.

I salute the desire to use a real database that, presumably, provides database integrity, since I have seen quite a few complaints in earlier versions about the iTunes database getting corrupted. But I question why the iTunes database has to be so damn slow. For all my complaints about the slowness of Mail, iT seems far worse, especially in the apparent random mindlessness of when it just times out for a minute or two doing god knows what, but not accepting any input.

I tried, in an act of bravery, to run this same SQLite vacuum command on the iTunes. sqlite3 told me that the database was in an encrypted format — which for all I know may mean what it says, that iTunes, for god knows what DRM reasons is encrypting the database, or may be the generic error SQLite throws up when given any file that isn’t a SQLite database.

So what’s the score here? Does iT use SQLite? Have they at least made an effort to use a real database engine that is reasonably fast, or did they just cobble together their own engine because god forbid they code like wimps and actually admit that someone else might be better able to solve a problem than themselves? Did they say “Core Data is nice, but we have to run on Windows, so screw you” rather than at least looking into what it would take to move the relevant parts of Core Data to Windows?