[OSM-dev] TIGER import.rb

Dave Hansen dave at sr71.net
Wed Jun 13 22:08:29 BST 2007


Cool!  I'm glad somebody else is interested in this stuff.  Did you get
code from the applications/utils/import/tiger directory in SVN, or did
you write it all yourself?

So, one thing I spent quite a bit of time on was making sure that all of
the ways that were adjacent and could be made into a single way were
merged.  This way, long lengths of streets don't get labeled many times.
This is a weakness of the renderer, but it makes for weird-looking
duplicated street labels:

http://labs.metacarta.com/osm/?lat=6163799.79449&lon=-13693354.33969&zoom=15&layers=B00

It also makes it hard if you want to modify the entire street and add an
osm tag, say highway=primary or something.

It actually wouldn't be too hard to merge our stuff together.  You could
pretty easily make your code do the same thing if you took my
create_way() and way.combine_with() functions and used them instead of
the simple addition to the way with 'way.segs << '.  One thing that I
completely ignored for now was merging tags, though.  We'd have to come
up with a way of reconciling them.

> Let's put our heads together, combine the features of our two
> importers,

Definitely.  One thing I saw in the data that I think is a bad fit for
OSM is the stuff like zip_begin/end or zip_left/right.  Shouldn't we be
tagging nodes themselves with zipcodes and addresses instead of the
ways?

> and work on some of the outstanding issues like the _incredible
> slowness_ of the upload. The county in the above map took about 10
> hours to upload. And it's a small, rural county. 

I thought that the original tiger import ran on the server and didn't
even go through the normal API.  But, that was before my time, and I
guess it was a bit of a fiasco.

I actually don't think 10 hours is too bad. :)

I think it would be really nice if all of the people interested in the
tiger data could use the same conversion programs to get the data into
OSM.  We're all going to have the same issues, and I would imagine we'll
need to come up with a few new tags and conventions for handling things.
Do you know of any other areas that are using the TIGER data?

-- Dave





More information about the dev mailing list