[OSM-dev] openstreetmapdata.com

Paul Norman penorman at mac.com
Mon Mar 11 07:02:42 UTC 2013


> From: Jochen Topf [mailto:jochen at remote.org]
> Sent: Sunday, March 10, 2013 11:42 PM
> Subject: Re: [OSM-dev] openstreetmapdata.com
> 
> This process is not and can not be perfect. If you wait for the
> coastline data to be perfect, you can wait for months. Every day errors
> are fixed but new errors are introduced. So it can take a long time and
> many runs of the OSMCoastline software till you happen to run it at a
> time when everything is "perfect". And of course that still doesn't mean
> the coastline is right, because we can only check for inconsistencies in
> the data and not against the real world.

Just for an idea of the number of errors, when I was actively watching it
was about 5 errors created and fixed a day, most of which would of not had a
noticeable impact on rendering in the area.

As a testament to how quickly the OSM community can fix coastline errors,
because I upload the small error file first I'd often find the errors fixed
before I loaded it up locally and downloaded the error, and I was generating
it locally!

This implies that the best time to generate an error free coastline is a
couple of hours after errors get published.

On my hardware I could generate the coastline every 60-90 minutes (io
limited for generating my coastline extract) but it would max out my
upstream connection. I think OSMCoastline is a bit faster, but really, once
a day is plenty.




More information about the dev mailing list