[OSM-dev] Map files for entire US (for download): use case for JOSM's layers?

Dave Hansen dave at sr71.net
Thu Jul 26 23:58:38 BST 2007


On Thu, 2007-07-26 at 17:09 -0500, Alex Mauer wrote:
> Dave Hansen wrote:
> > I'm uploading a test county now (Hood River, OR) so that we can see how
> > it renders.  Please don't upload any other bits of this, just yet.  But,
> > please download them and take a look at them in JOSM or even render them
> > if you can.
> 
> What problems should we look for?

Streets that should be there and aren't.  Nodes that are shared between
a power line and a road.  Streets with the wrong names.  Even connected
ways that have the same name would be errors.  The script should have
combined those.

> And if no problems are found, why not
> upload?  At the very worst, it's still better to have some data to start
> from (and correct problems) than to have to start from scratch.

Mostly because if there is a problem, it is somewhat time-consuming to
delete an entire county and re-upload it.  Especially if there is a
problem that affects the entire data set.

> It also occurs to me that this would be a great use case for JOSM's new
> layers when merging TIGER and OSM info where it already exists: Load the
> OSM data; load the imported TIGER data; select a way in the TIGER data;
> if JOSM can do this yet, copy or move the way from the TIGER layer to
> the OSM layer.  et voila, the TIGER way exists in OSM.

That would be nice.  For now, I'm just planning to upload them, and fix
up the duplicate streets that I have in my area.  I haven't done _that_
much work.

> It would also be wonderful if JOSM can select everything with a certain
> tag for the copying: I have (almost) no rail whatsoever in my county,
> like 200' or so...so little that deleting it and importing the TIGER
> data is very feasible.  But even assuming the inter-layer copying (as
> above) is doable, can I select "all rail" or do I have to select every
> single individual railway way?

I could modify the scripts to just spit out rail data.  That wouldn't be
too hard.

-- Dave





More information about the dev mailing list