Question

I’ve got what should be a pretty simple CoreData setup that looks like this:

Track {
    NSSet *artists;
    NSSet *genres;
}

Artist {
    NSSet *tracks;
}

Genre {
    NSSet *tracks;
}

So the idea here is, a Track can have multiple Genres and Artists. There is an inverse relationship back to Track in both cases.

I’m trying to get a list of Artists which have at least one Track of a given Genre. I’m using the following predicate: [NSPredicate predicateWithFormat:@"ANY tracks in %@", genre.tracks].

Most of the time this is fast, but sometimes I have a Genre with 10k+ Tracks. This generates a SQL query with 10k+ variables… this takes a long time to run.

I’ve tried all kinds of things to avoid this including:

[NSPredicate predicateWithFormat:@"ANY %@ in tracks", genre.tracks]
[NSPredicate predicateWithFormat:@"SELF IN SUBQUERY(tracks, $t, %@ IN $t.genres)", genre]
[NSPredicate predicateWithFormat:@"(SUBQUERY(tracks, $t, %@ IN $t.genres).@count > 0)", genre];

And probably some more that I’ve forgotten. Each of these compiles, but returns no Artist objects.

What can I do to improve upon the efficiency of the IN query I started with (which works)?

Answer 1

You can use

[NSPredicate predicateWithFormat:@"(SUBQUERY(tracks, $t, ANY t.genres = %@).@count > 0)", genre]

which gives the same result as your original predicate, but avoids to build a query with many parameters.