User talk:Nick SPW

Salisbury, Maryland import

Started from talk-us@ mailing list discussion Uploads to City of Salisbury, MD about problematic import. Here is my first pass at the changesets:

(Format this)

10882159: 493 dangling nodes (reverted in 11069425)

10884039: 3517 dangling nodes (reverted in 11069447)

10885267: 1 building node (OK)

10891857: 5 dangling nodes (reverted in 11069488)

10891870: 8000 dangling nodes (reverted in 11069513)

10893364: duplicate ways (REVERT)

10894586: relations mistagged (REVERT)

10899769: bad tags,no building=yes (REVERT)

10900393: bad tags,no building=yes (REVERT)

10901301: 20000 dangling nodes (REVERT)

10901673: seems ok? but multipolygon members are also tagged building=yes

10902224: seems ok? but multipolygon members are also tagged building=yes

10903083: seems ok? but multipolygon members are also tagged building=yes

10904229: seems ok? but multipolygon members are also tagged building=yes

10909376: seems ok? but multipolygon members are also tagged building=yes

10910137: seems ok? but multipolygon members are also tagged building=yes

10911082: seems ok? but multipolygon members are also tagged building=yes

10911716: seems ok? but multipolygon members are also tagged building=yes

10913242: seems ok? but multipolygon members are also tagged building=yes

Major question is if #10901673 and on imported the same data, or if everything prior to that was just testing? That might save us from having to revert everything... -- Joshdoe 11:24, 22 March 2012 (UTC)

I did do a lot of testing within JOSM before I actually figured out how to upload the building footprints effectively. I made attempts to import the entire dataset, which was too resource intensive. When the imports started to become consistent was probably when I realized I needed to break the building footprints into 10 10% "percentiles" then upload incrementally.--Nick SPW 12:56, 22 March 2012 (UTC)

For future reference there are development APIs for just this purpose, for testing clients/methods/etc. So is the entire Salisbury building dataset uploaded? And how did you split them up? You mention 10 "10-percentiles", but I only see 9 seemingly consistent changesets, so I'm not sure if you stopped early. Another question is if you have address data that can be correlated to the buildings prior to import. I don't intend to fragment this discussion, but ... -- Joshdoe 13:08, 22 March 2012 (UTC)

Yes, the entire Salisbury metrocore building dataset has been uploaded. To split the dataset, I used Manifold GIS to run SQL SELECT TOP 10% and SKIP x% and exported each result to .shp, resulting in 10 equally sized .shp files. Changesets 10899769, 10900393, 10901301 may have been attempts and failures to upload my first 10%. I do have address data that can be correlated to the buildings, but the data is kept separate by the stewards to deal with multi-unit buildings. I would need to plan how to deal with multi-units in JOSM to include addresses. If addressing is something beneficial to pursue, I would also be concerned with replacing the 2007 TIGER data with our higher-accuracy street centerline data. --Nick SPW 15:22, 22 March 2012 (UTC)