benchmarks about vtquery HOT 2 CLOSED

@mapsam - I would be very interested, once the basic implementation is working, for you to benchmark rough performance against node-mapnik's vtile.query.

The reason is that the results I think should help guide level of effort and potentially what related pushes we invest in.

For example if the speed is similar, then I'd be concerned that closest_point is expensive (we'd of course confirm with profiling) and we'll then want to prioritize investment in trying contains, within, or covers instead per #13 (comment) (when radius=0). Or potentially looking at what is slow inside closest_point and trying to optimize it.

If the speed is faster, then that would be exciting since we'd then know we have room to do more while keeping the current API latency similar: like querying more surrounding tiles to support larger radius values without needing to jump to parent tiles (that might not contain all the features).

Note: Since node-mapnik only takes one tile, we'd stick to one tile for 🍏 to 🍏.

from vtquery.

Benchmarks have been set up and are now helping us make decisions when profiling and making updates!

They can be run like so:

node bench/vtquery.bench.js --iterations=1000 --concurrency=1

from vtquery.