Skip to Content.
Sympa Menu

overpass - Re: [overpass] Query by quad tile key

Subject: Overpass API developpement

List archive

Re: [overpass] Query by quad tile key


Chronological Thread 
  • From: Jan Marsch <>
  • To:
  • Subject: Re: [overpass] Query by quad tile key
  • Date: Wed, 22 Feb 2017 16:08:31 +0100

Hi Roland,

I've had clearly no insight into the current structure. My assumption was
something like this: https://wiki.openstreetmap.org/wiki/QuadTiles
And I'm missing even more understanding if this is maybe exactly what you are
describing with bbox limitations.

If it's good enough to address features in a z/x/y schema translated to a
bbox then great.
I'm doing this already on client side but this is then like any other bbox
for Overpass.

For features crossing boundaries, including them as a whole would be fine.
Any other behavior (filtering, clipping) could be done on client side.

Regards,
Jan

> Am 16.02.2017 um 07:05 schrieb Roland Olbricht <>:
>
> Good morning,
>
>> Is there a way to query by just providing a quad tile key?
>> Further options are not required. Just this.
>> For background: I need all data and have rather complex processing in mind.
>> Result will be tile based anyway and data structure seems already very
>> suitable for that.
>
> Not really. The problem is that the ways are in overlapping index
> structures. Hence, a strict quad tile based approach would miss those ways
> that cross a tile's border.
>
> What works now is to have bbox limitations
> from 0.0000000 + n * 0.0065536 to 0.0065535 + n * 0.0065536
> This matches exactly the tile boundaries and assures that each way is in at
> least one tile.
>
> What also works is to use "out qt" (or "out meta qt"). This orders ways by
> quadtile and is guaranteed to produce each way exactly once.
>
> To have a quad tile criterion I would need to know what should happen to
> boundary crossing ways. If you just need to get data in chunks then I would
> then I could make a filter "chunk" or so that gets a integer, without a
> guarantee that ways are spatially ordered but with a guarantee that each
> way is contained in exactly one chunk. The chunks will be of very different
> sizes.
>
> Regards,
> Roland
>




Archive powered by MHonArc 2.6.18.

Top of Page