[OSM-dev] Data Consistency Checks

[Harald Kleiner]

Hi Bernd! 

So I'm sorry to admit, but you were first!  ;-)

loose ended ways seem to be a very common and hard to find problem...

> In January I have written some test code for "Unconnected ways"...
> 
> The test code is not really efficient because it starts with
> collecting all ways...

Well, I found a similar algorithm independently. So it cannot be that bad!
It is surely a hard task because you have to check many distances to all ways in the near environment

My scope is always the complete database (which contains just Austria)
Have you ever tried to run your check on an area "as large as" Austria? Would be interesting, if java can keep up with plain old SQL...

A few words about my method:
* Find end-nodes of ways that are not connected to any other way and check for each of them:
* find ways that have at least one node inside the 3x3 matrix of tiles surrounding the node-under-test
* calculate the min. distance from the node-under-test to any segment of these ways
* if the min. distance is less than a given limit, an error is found (the node lies almost on a segment)

The last part is the really expensive one...


lG Harald
-- 
GMX startet ShortView.de. Hier findest Du Leute mit Deinen Interessen!
Jetzt dabei sein: http://www.shortview.de/?mc=sv_ext_mf@gmx