[Imports-us] Union County Original Data and OSM files

[Paul Norman]

> From: Randal Hale [mailto:rjhale at northrivergeographic.com]
> Sent: Sunday, December 08, 2013 12:21 PM
> To: Imports US
> Subject: [Imports-us] Union County Original Data and OSM files
> 
> I placed two links here:
> http://wiki.openstreetmap.org/wiki/Data_Import_for_Union_County_Georgia

Thanks for the files.

I had a look primarily at the and had a few comments

- The tags are addr:housenumber, addr:street, etc, not address:* 

- Housenumbers with a letter (e.g. "143 a") have a space and are 
  lower-case. This might be correct, but you should check, I've seen a lot 
  more "143A" in the field. 

- You're probably going to have to filter out every housenumber not 
  beginning with a number, they all appear to be notes. "burned down" 
  should probably be used to indicate manual checking is required, of the 
  two occurances in the sample, one doesn't correspond to anything on the 
  ground and the other shows a ruin. demolished:building would be my 
  guess, but I'm not a tagging expert on those. - You're going to have to 
  do some cleanup on the street tag. <null> <null> is the most obvious 

- Besides Dr, Mtn, E and Rdg are also unexpanded 

- I happened across these street values that seemed odd and should be 
  checked: "Z 28 Dr", "E.a.s.t. Point", "Davis Dr (fannin)" 

- Buildings on FS roads and numbered highways don't have a street 
  tagged. Probably a source data issue, but worth noting. 

- Are you selecting building=* values from an explicit list and falling 
  back to building=yes for unknown values? 

- Don't add address tags to objects that don't have one in the data. 
  Right now every shed has them, even if it doesn't have a housenumber or 
  street. 

- The building in building errors are definitely from the source data. 
  There are also some geometry errors with self-intersecting buildings, 
  etc. I can't see that these can be fixed automatically. The best 
  approach would be getting them to clean up the geometry and building in 
  building errors in their data. I could generate a list of OBJECTIDs 
  which are involved in errors if it would help.

Overall I'd say it looks reasonable for a first review. As the maintainer of
ogr2osm I'd be interested in looking at the translation file used.