[OSM-dev] GRR!

Ben Gimpert ben at somethingmodern.com
Wed Apr 12 10:34:53 BST 2006

So for the last week-or-so, I've devoted my OSM lurking time to actually
writing some code.  (Hence I've been even more remiss on the mailing
list discussions.)  Steve asked that I update my very stale TIGER import
code for the new 0.3 REST API, and kick off the import for the whole of
the US.  'Tis done, and just committed:


My priorities have been good exception handling an exponential server
backoff mechanism, since this import will certainly peg the OSM server.

Unfortunately, unless someone is inspired to optimize my grotty street
merging code, I will disable it on the import.  This means that the 15
perfectly contiguous *but separate* streets called "Lake Shore Drive" in
Chicago will end up in OSM as 15 perfectly contiguous ways...  Such is
the quality of the TIGER dataset.  Maybe that's why this is a wiki,
right? ;)

To kick off the import, I need a proper Linux and Ruby environment.  The
machine must be up 24/7 (which knocks out my home laptop), and it must
have at least 10GB of space (which knocks out my lovely Bytemark VM).
The import will probably run for a week-or-so, but should not tax the
machine much.  Any volunteers?


More information about the dev mailing list