Question

I have a pretty simple little schema that looks like this:

Trail –> Segment –> Coordinate

All of the proper relationships are in place to make this work, and they certainly let me do something like this:

NSArray *trails = <query all trails>

for (Trail *trail in trails) {
    for (Segment *segment in trail.segments) {
        for (Coordinate *coordinate in segment.coordinates) {
            //do something with coords
        }
    }
}

Coming from a Django background, I was delighted with this, and assumed the fewest possible queries would be run against sqlite3. Wrong! There is a query made for each and every object in these loops.

I’ve now read the Core Data documentation on faulting, I understand what’s going on, although I would really like to know what the best practices are for dealing with a situation like this. I need every single object in memory (the coords model has ~100,000 objects) because they are all used at one moment to plot a trail on a map.

How should I deal with this ultra lazy loading?

Answer 1

You could use the setRelationshipKeyPathsForPrefetching: method of NSFetchRequest for this. From the documentation:

Prefetching allows Core Data to obtain related objects in a single
fetch (per entity), rather than incurring subsequent access to the
store for each individual record as their faults are tripped.

So in your initial fetch request for the trails you could add

[request setRelationshipKeyPathsForPrefetching:
    [NSArray arrayWithObjects:@"segments", @"segments.coordinates", nil]]

(I never tried this with 2 levels deep, only with one level.)