Incorrect road labels? (Was: [OSM-talk] Freemap in Mercator)

[Christopher Schmidt]

On Tue, Jul 04, 2006 at 07:45:37PM +0200, Petter Reinholdtsen wrote:
> 
> There are something wrong with the map data being displayed, though.
> 
> Check
> <URL:http://london.freemap.in/july.mercator.html?lat=10525093.625&lon=1852014.67969&zoom=10>.
> The Lænsvæg 131 and Riksvæg 50 roads are not correctly labeled.  Those
> roads are in sweeden, not in Norway.  Anyone know how this could
> happen?

http://www.openstreetmap.org/api/0.3/way/1736075 seems to be the way in
question. This has, at the top, <seg id="483"/>, which is:

 <segment id="483" from="492" to="495"/>

<node id="492" lat="68.3909072875977" lon="16.5840244293213"/>
<node id="483" lat="68.3943786621094" lon="16.5887966156006"/>

This leads me to believe that the segment 483 should actually be
something near to:

http://www.openstreetmap.org/api/0.3/segment/483636
(which is also in that way):
  <segment id="483636" from="536916" to="536928"/>
  <node id="536916" lat="58.0446370075175" lon="14.9737429618835"/>
  <node id="536928" lat="58.0439783155355" lon="14.9749231338501"/>

So, the 'label' there is probably being displayed on the single segment
483, which is wrong in the way, which should actually be pointing to
483***, and this is as a result of way API calls getting 'chopped off'
when being posted to the API. I mentioned this in the data last month --
Federal Highway, in Australia, went all the way into England becasue
Federal Highway was 'chopped off' after a certain number of segments.

This has been the case for other people posting 'long' XML data to the
API for a while now, and has been mentioned on #osm on IRC before.

I'm not sure if this is a known issue -- I recognize it, but haven't
looked into the code to see why it's happening. It seems likely that
either the Applet, JOSM, or API are not dealing with long API calls.

Turning off segments for the map you linked to shows that those labels
are only on one segment of the map, which seems to further accentuate
why this is happening.

Does that explain what you're seeing?

In the first planet.osm dump, someone offered a diff to the planet.osm
file which found these things, and fixed them, but I think it might have
been manually created. If someone can offer the same for the current
dump, I'll apply it and clear the cache so these segments go away.


-- 
Christopher Schmidt
Web Developer