ratrun

> PreparedPolygon.intersects(Geometry) has an optimization if the polygon itself is a rectangle. So you would need to create a prepared version of your bbox polygon and switch the order in...

> Can we take a step back and think about what we want to use landuse for? I would like to use `landuse` for greenery bicycle routing. There is an...

> If a route avoids residential landuse, wouldn't it be considered greenery? Basically yes. But I expect problems with `landuse=grass`, this and the following list also needs to be considered...

> Why is landuse=grass not 'greenery'? According to the [wiki](https://wiki.openstreetmap.org/wiki/Tag:landuse%3Dgrass) it is to be used for small areas. From what I have seen so far it is also used most...

> That should work fairly well using urban_density You are probably right. But with the landuse encoded value it would also be possible to prefer `forest` with a higher priority...

> Ok so besides urban_density we would only need landuse=forest? I think that this would be sufficient, but other people might have other needs.

This branch looks very promising, although it is quite expensive in terms of RAM and preparation time. I tested the extract for Austria with four bicycle profiles active and had...

I created PR #2812 which implements the mentioned search buffer from above. @easbar Can you please check?

> Querying with a buffer like you proposed should fix this. Yes, it does: 

We rank `highway=secondary` slightly down even if these are tagged with `sidewalk=both` and the maxspeed is smaller compared to 50 km/h. See https://github.com/graphhopper/graphhopper/blob/fe2ce71b3c96f6db65a2d0856a6d39c229c5a440/core/src/main/java/com/graphhopper/routing/util/parsers/FootPriorityParser.java#L120 And we rank up `highway=residential` and `highway=footway`...