[OSM-dev] Problems with TIGER Conversion

Al Wold alwold at gmail.com
Thu Jul 12 18:52:20 BST 2007


On 7/12/07, Dave Hansen <dave at sr71.net> wrote:
>
> On Thu, 2007-07-12 at 02:32 -0700, Al Wold wrote:
> > Dave,
> > It looks like I'm getting missing data (I think mainly entire ways are
> > missing) with the 0.3 version.  I had a file that seemed more complete
> > from an earlier version, which I think was 0.2.  I'm regenerating with
> > 0.2 and I'll see how it looks when it finishes.  If you want more
> > details on what is missing, let me know.  If you don't have time to
> > look into it, I may have some time to dig into the code tomorrow.
>
> Gah.  I'm noticing the same thing with larger files.  My own street is
> messed up. :)


It looks like the 0.2 script didn't output any tags at all on the entities
it output.  Weird, I wonder which version I was using before.

> The stuff from the earlier version was looking pretty good. I did
> > notice that it doesn't seem to incorporate the address ranges.  I
> > think you said earlier that the original tiger attributes were being
> > carried over so it could be done later, but I don't see them in either
> > the ways or segments.  It probably has to be tagged in the segments
> > right?
>
> The nodes and ways have their original tiger attributes in them.  That
> means that we can find an individual, original tiger feature, given a
> node or way.


I don't see any of the FRADDL, TOADDL, FRADDR, TOADDR fields in the result
file.  What attribute would they be mapped to?  (I searched for "addr" and
"ADDR" and found nothing)

> I've got the ruby script to create excerpts from the converted files
> > to a decent state, and if you use it to cut the file into decent sized
> > blocks, JOSM is able to deal with them without any trouble.  I've
> > attached the script for anyone that wants to try it out.
>
> Hmm.  Perhaps we should integrate this into the actual converter.  Were
> there any tricky bits that you ran into?


Nothing too tricky.  I ended up doing the lazy method and assuming
everything was defined before being referenced, and I just cut off ways at
the edge of the bounding box.  I'm not sure if integrating the two scripts
would help that much.  You'd still have to process the entirety of the tiger
files for the county wouldn't you?  So, that would still take just as long.
Maybe I'm nothing thinking of something though...

-- Dave
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/dev/attachments/20070712/def9b3ea/attachment.html>


More information about the dev mailing list