User talk:Manchito

From OpenStreetMap Wiki
Jump to: navigation, search

Nodes not in ways (Colombia)

Hello. I'm finding that there are lots of nodes which you have uploaded which are free standing, not a part of a way. Is this what you have intended to do, or is it an unintended side-effect? An example, the way illustrated in has points uploaded by me in the context of the way; those points are identically placed to points you uploaded as free-standing nodes. In the view linked above, if you look at the Data overlay you can see both the free standing nodes and the way. Thanks for your thoughts on this. --Ceyockey 16:43, 23 December 2010 (UTC)

P.S. the same area in the DupNode resource → --Ceyockey 16:44, 23 December 2010 (UTC)
Thanks for asking. Those nearly 50.000 isolated nodes are the result of an incomplete import of the first 10 cluster files. Due to API restrictions, the upload had to be performed in several changesets. 95% of the process went well, but it lasted so long that, at the end, some users had already deleted some of the nodes, which broke the last step: the import of ways and relations, thus leaving all those orphan nodes you see. There were just too many conflicts, all to be solved in the same way, but AFAIK JOSM has no such en masse conflict-solving feature. (Besides, the API was awfully slow!) I discussed this with Mike and he agreed to look after that and reprocess my file as soon as he had tweaked a script of his. I can see he started with the first two, don't know what happened with the rest. Meanwhile, and due to the urgent need of data on some areas, I'm re-importing some of those files making sure not to duplicate nodes. I suppose you could give us a hand with that too. I'm open to suggestions. --Mancho 20:33, 23 December 2010 (UTC)
I had been deleting some when I found them, and it sounds like either deletion or merger would be OK based on your description. My target has been cleanup related to my own relation loads where I've inadvertantly introduced or have found in proximity significant amounts of redundancy. Given my relatively small geographical focus in any one particular session, it would appear the best course for me to help might be to delete by the following method: a) select all within a JOSM session; b) search-and-select from selection all elements belonging to changeset 6546491 (that's one of the special search functions in JOSM); c) delete all then selected elements. Would that be an acceptable solution, or is it too scorched-earth? It sounds like you are looking at a more fine-tuned solution. --Ceyockey 22:54, 23 December 2010 (UTC)

I've started doing some 'minimalist' boundary additions with Margarita. What this means is that I only import information needed to make the boundary; if pieces of the boundary already exist, then I use those. In the case of Way 91508608 (XML, Potlatch2, iD, JOSM, history), your points were already present, so I just connected these to create a way which is part of the Margarita municipality relation. It takes as much time to clean up after myself when blind uploading as it does to do it this way, and it is a major reduction in the actual amount of data added to OSM and the upload traffic. This is only possible due to prior work leading up to the current state, though. P.S. I'm not done with this particular relation, which is why it remains open as of this writing. --Ceyockey 22:48, 24 December 2010 (UTC)