Note that you can also use the main site for that now (the bigger map is linked from the relation browser). Although the external site may look better.

randomjunk: In this particular case this seems to have been a prank by a sockpuppet of an experienced OSM user, as witnessed by - among other things - a circular 12 nautical mile administrative zone, which incidentally is still on the map.

But generally speaking it's good to ask people to fix things they get wrong themselves, even if that means waiting a bit more. The resulting learning experience is often very useful for new contributors and often avoids hurt feelings people may have when they see their work deleted. Even if it doesn't align with the goals of the project.

It's little stuff like that which helps to maintain a friendlier atmosphere than the (English) Wikipedia.

This is actually the first bit of Windows programming I've ever done, and I'm doing it in AutoIt which is a language I've never used before. But it's really neat, I wish there was something like it for *nix.

I'll try to clean up what I have and get it into SVN in a few days.

It's like that because a waterway=riverbank area is missing in places.

It also needs cleanup for other reasons, for instance the ways for the suburbs Lehel and Haidhausen don't share nodes at their edges.

Actually not only for the user edits but also for recent changes and history within a bbox now. See the ticket in trac.

What encoding is the data originally in and how are you converting it? You converting "cour à bois" to "cour [U+FFFD] bois" suggests a easy-to-fix encoding issue in your code.

And yes, structured data is better, so perhaps canvec:value=Lumber yard - ( Cour à bois ) should be split up into canvec:value:en=Lumber yard and canvec:value:fr=Cour à bois. Although that tagging schema suggests those values came from the canvec dataset.

Your use of the U+FFFD replacement character in the data suggests some sort of encoding conversion problem.

I think it looks good, not too many tags (if there's such a thing).

You can remove point clouds with gpsbabel.

And even if you don't make the traces public they'll still appear in the GPS API, so someone requesting your area might still get your point clouds.

It's very nice. But has some bugs, mostly with the various pop-up dialogs tripping over each other.

Might I suggest you include Google MapMaker as well?

Ah, src/osm/sites/tile.openstreetmap.org/cgi-bin/export

Where's the source for the cgi-bin/export script that does the mapnik exports?

This blog posting is being discussed here.

If you have a bunch of .img files you can combine them into a gmapsupp with:

java -jar mkgmap.jar --gmapsupp list.img of.img files.img

Also, to produce this without splitting in the first place you can add more virtual memory to your machine, it'll be slower but at least it won't run out of memory instead of finishing slowly .. eventually.

Have you tried using splitter.jar to split up denmark before you make the map with mkgmap?

There's no reason why you need to make one giant .img file of Denmark, and if you split it up you could make it on your computer without stripping any data from the .osm file.

Presumably because it comes from the TIGER import which is notoriously inaccurate.

Looks like you found the signup page Ciarán. I just put my name there and had someone contact me via e-mail.

If it's a shapefile or in another similar GIS format you could import it directly into OSM as-is with a conversion script.

There's a shp2osm script in the OSM svn, but you should ask on the OSM talk mailing list and provide links to the specific data you're working with, there are people there experienced in importing that can give you good advice.

OpenPisteMaps renders it along with osmarender.

Nice I added a relation describing Hlöðuvallavegur and Skjaldbreiðarvegur, added them to the list of national highways on the Icelandic Wikipedia and added appropriate ref= tags based on vegaskrá.