Investigate data query speed enhancements.

[watmildon]

We keep adding more data and it's making the map load time slower and slower. Some ideas:

• Add a QLever query as one of the style file options, loads faster at the expense of only updating once a week
• Generate a geojson file every few hours in a github action as a cache file, more complex but should be fast to load and more quickly up to date

Any other options that don't involve me running a SQL database and monitoring all changes? lol

[GA-Kevin-Codes]

Do we know what is taking the longest? Is the limiting factor the device you use or the server the query is run on?

[watmildon]

It's Overpass returning the query. Some rough benchmarks from my machine just now:

nwr[mascot][amenity=school] - about 20 seconds to load the USA
nwr[mascot][amenity=school][name~"High School"] - about 30 seconds to load the USA

Anecdotally, QLever queries of this size are usually 5-10x faster.

The rendering doesn't take too long but the experience could be improved (you can actually see all the icons pop in on top of one another when it loads).

[watmildon]

This query returns the ~100 items that have been in the DB for a long time. It will return more in ... a week?... whenever they update their side. It returns instantly right now.

It does return a full set of OSM tags yet which I would like to have for the map. I'll keep playing with it.

---
title: Query mascots on OSM using QLever
description: Load OSM mascot data from the QLever osm-planet dataset.
type: sparql
server: https://qlever.cs.uni-freiburg.de/api/osm-planet
options:
  zoom: -0.1
  center: [0, 30]
---

PREFIX geo: <http://www.opengis.net/ont/geosparql#>
PREFIX osmkey: <https://www.openstreetmap.org/wiki/Key:>
SELECT * WHERE {
  ?osm_id osmkey:amenity "school".
  ?osm_id osmkey:mascot [].
  ?osm_id geo:hasGeometry/geo:asWKT ?geometry
}