[OSM-talk] TIGER Addressing Import

Hillsman, Edward hillsman at cutr.usf.edu
Tue Oct 13 18:37:17 BST 2009


On Tue, Oct 13, 2009 at 18:20, SteveC <steve at asklater.com> wrote:
>
>Dave - super awesome.
>
>As I said on IRC the other week, but I'll repeat here for all - I  
>think dumping the addressing for all 3,000 counties and then letting  
>people import them one by one will be the best way to do it.
>
>Another random thought - should the addressing ways be one long way  
>with two nodes per block, or lots of two node ways? My immediate  
>preference is for the former...?

Although my preference would also be for the former, people will break
the ways midblock, for very good reasons. For example, recording a small
bridge in the middle of the block would entail splitting the way into
three--one for the bridge, and one on each end of it. Most likely, each
piece will still carry the full address range, even though the bridge
should have none and the range should be split. In Tampa, we also have
instances of streets transitioning from 4-lane undivided to 4-lane
divided ("dual carriageways") midblock.

Which raises another issue. The old TIGER files had a single way for
most of the dual carriageway major streets in our area. I have not
looked at the new 2009 TIGER files, but I suspect that they have similar
representations of these streets. When splitting these into two parallel
one-way routes, should both then carry the original addressing ranges,
or should one carry the odd address numbers and the other carry the even
address numbers? 
>
>Also - the ways will be deplaced 90 degress to the road centerline to  
>push them to the edge of the road I assume - but you also need to  
>'pull in' the end nodes too so they are not laying on top of the cross

>streets at each end, if you see what I mean?
>
>Yours &c.
>
>Steve





More information about the talk mailing list