See https://github.com/openplans/openblock/pull/4 for full explanation.
Excerpt:
"""
I believe I've uncovered a small bug with the OpenBlock block importer. TLIDs (from the edges shapefile) can have more than one associated feature name (primary and alternates). For example, the TLID for a segment of W. Franklin (in Chapel Hill, NC) has two names: W. Franklin (primary) and 54 (alternate). Here's an example dump of feature names matching TLID 72390848:
Notice the PAFLAG attribute on each record: P(rimary) and A(lternate). The OpenBlock block importer ignores this flag and ends up storing the last feature name that it finds during importing. This means that depending on the order of the feature name data per each TLID, certain blocks may be labeled with different names along the same street. Like "100-200 Franklin St." and "201-300 54".
"""
The pull request partially addresses this, but the comment describes outstanding problems with it that need to be addressed.
See https://github.com/openplans/openblock/pull/4 for full explanation. Excerpt: """ I believe I've uncovered a small bug with the OpenBlock block importer. TLIDs (from the edges shapefile) can have more than one associated feature name (primary and alternates). For example, the TLID for a segment of W. Franklin (in Chapel Hill, NC) has two names: W. Franklin (primary) and 54 (alternate). Here's an example dump of feature names matching TLID 72390848:
https://gist.github.com/1560992/5e07f01ad2745411059886347db73879c44de7cc
Notice the PAFLAG attribute on each record: P(rimary) and A(lternate). The OpenBlock block importer ignores this flag and ends up storing the last feature name that it finds during importing. This means that depending on the order of the feature name data per each TLID, certain blocks may be labeled with different names along the same street. Like "100-200 Franklin St." and "201-300 54". """
The pull request partially addresses this, but the comment describes outstanding problems with it that need to be addressed.