[OSM-talk] What's wrong with this picture?

Michal Migurski mike at stamen.com
Tue Dec 13 06:33:47 GMT 2011


On Dec 12, 2011, at 8:38 AM, David Groom wrote:

>> I recently made some updates to the code behind the Metro Extracts (http://metro.teczno.com) that pushes the coastline files through PostGIS and hunts down additional errors there. Many aren't interesting (nested holes) but some do lead to bugs in PostGIS and Mapnik's ability to render tiles, self-intersections being the largest problem. I'll be updating the site with new error shapefiles when the process stops running over the next 24 hours.
> 
> It will be interesting to see how many errors that highlights

A few sample errors:
	http://tile.stamen.com/coastline-next/preview.html#7/46.844/-86.764

The two pink "2's" were identified by the coastline error checker, and (I think) indicate an island with part of its coastline facing away from the water. The three green "X's" represent self-intersections picked up by PostGIS, and if you zoom in more closely you'll see where they are to fix them. I've recently fixed up a number of self-intersections around South America, Mexico and the US:
	http://www.openstreetmap.org/user/migurski/edits

The data for this view is linked here:
	http://metro.teczno.com/#coastline

…and the stylesheets are here:
	https://github.com/migurski/Extractotron/tree/master/coastline-preview

I hope to keep this view up to date regularly, though it's possible I'll forget from time to time. I've been watching edits near these points and there are a few mappers getting to them before I do:
	http://www.openstreetmap.org/user/dmgroom_ct/edits
	http://www.openstreetmap.org/user/PA94/edits

I've been dating blocks from versions of the coastline, they can be viewed here:
	http://tile.stamen.com/coastline-current/preview.html#8/46.106/-84.303

-mike.

----------------------------------------------------------------
michal migurski- mike at stamen.com
                 415.558.1610






More information about the talk mailing list