I’m new to OpenStreetMaps, but I’ve been researching/experimenting it extensively over the past weeks, and as a result have a few questions: I’ll post just one here for simplicity’s sake ; )
My work concerns the City of Paris (France), and I’ve noticed that contributors there have given the “building” key only to the outer polygon in multipolygon buildings (meaning that, should a building have a ‘hole’ in it, the hole is ‘related’ to its multipolygon through its id number, but has no tags itself). This causes problems in some rendering cases, and I see that the OSM wiki says that, should a multipolygon ‘object’ have several members, it should be the ‘multipolygon’ attribute that gets the tags and not one of its members.
So what I would like to do is a) find all ways with the ‘building’ key and b) check to see if they are members of a multipolygon and, if so, c) move the tags from the ‘member way’ to the multipolygon itself. Unfortunately, my target area has several thousand cases of this sort within.
I can think of a method of doing this through scripting (compiling a modified OSM file from an OSM original, through php or perl, that I would then re-upload), but perhaps there already exists a tool that does this?
I believe you’d get better answers from the OSM dev mailing list (dev@openstreetmap.org)
All in all, OSM does not really have a proper support for polygons. There is an open but quiet discussion going on at http://wiki.openstreetmap.org/wiki/The_Future_of_Areas. Please read and add your comments and suggestions.
Thanks for the insight ; ) But mailing lists are a messy lot to follow ; P
It is entirely possible to implement this sort of idea, but it must be done within hours - I have the osmosis/php scripts already in place, but the ‘bulk’ transformation must be done as quickly as possible so to avoid any disruption in the (constantly) updating process. I’ve already asked questions of my own about this method of treating what should be ‘object’ data - if you like, you can have a look here: http://www.openstreetmap.org/user/ThePromenader/diary/13561.
PS: don’t worry: I’m only working on the area I’m concerned with - Paris, France - and I won’t be doing a thing until I am 100% sure that it is working correctly, in a timely fashion. Cheers.
As for the ‘The Future of Areas’ wiki - an interesting read, but there doesn’t seem to be much discussion there - I made a short post (hoping to open a discussion elsewhere) but it was eliminated outright. I guess they want to ~keep~ the discussion quiet.
Instead I’ll try the mailing list - I’ll make a general proposition and see if that generates any interest.
Automated edits & bulk changes are known from bitter experience to cause problems. I presume much of the data you are talking about is derived from the Cadastre: this is imported data, often imported in haste and not always checked over. I would not regard cadastre derived data as an example of good practice in OSM.
You yourself have only recently become involved in OSM, and carrying out such edits in an area which is as intensively mapped as Paris is highly inadvisable. Furthermore your proposed edit is to merely fix a rendering issue, rather than enhancing the data.
At the very least discuss this on talk-fr, and be aware that efforts to clean this data up in the past have not been well received: http://www.openstreetmap.org/user/Pieren/diary/12250. See other diary entries by this user who has spent time correcting data in Paris.