Changeset: 31269952
Added.
Closed by Paul Cone
Tags
browser | Firefox 38.0 |
---|---|
created_by | iD 1.7.2 |
host | http://www.openstreetmap.org/id |
imagery_used | Bing |
locale | en-US |
platform | Windows |
Discussion
-
Comment from Peter Dobratz
Hi and welcome to OpenStreetMap. You placed a Node in the middle of a building here with tags on it for the A-Boy hardware store, but it turns out this information is already tagged on the building outline. Generally putting the tags on the building outline is considered the preferred approach and considered a higher level of detail. In any case, we want to avoid duplicate data where possible, see http://wiki.openstreetmap.org/wiki/One_feature,_one_OSM_element
Let me know if you have any questions. -
Comment from Paul Cone
Peter, the problem with that approach is if you need to use a derivative of the polygon dataset for other purposes, you will not get everything you need, i.e. because not ALL of the hardware stores have polygons, one must use the point data, too. I understand that having the feature only once is the preferred approach, but until the data catches up with the policy, I think it's better to have both a point and a polygon rather than have to use both and then go through the task of weeding out duplicates.
-
Comment from Peter Dobratz
Yes, there will be some hardware stores in OSM as single Nodes and some as Closed-Way polygons. Additionally, there are Relation objects of type multipolygon which might be hardware stores (buildings with a courtyard or multiple buildings). The shop=hardware and contact info tags should be on either the Node, the Closed-Way, or the multipolygon Relation. It's bad practice to put the tags on more than one of these 3 options.
We shouldn't be adding duplicate data into OSM on purpose just to facilitate the needs of a particular data consumer. You should be able to convert polygons to points to facilitate your processing of the data.
By putting a Node with shop=hardware inside a Closed-Way with shop=hardware, you are basically saying that you have a hardware store contained within a hardware store, which is incorrect. In reality, there is only one hardware store here.
-
Comment from Paul Cone
As I said, I'm quite clear that I can convert polygons to points, but then how would you suggest I weed out the duplicates between the points derived from polygons and the points that orginally were points?
-
Comment from Peter Dobratz
We shouldn't be adding duplicate data to the main OSM database on purpose. The "polygon dataset" you speak of is not something that is native the OSM data model, but something that was derived from OSM data. We shouldn't require duplicate data in the OSM database in order to help you weed out duplicate data in the dataset that you have derived from OSM data. Your application should be able to consider OSM data as a whole (Nodes, Ways, Relations). Is it worth posting this on an email list to get some more people to weigh in?
-
Comment from Paul Cone
If the polygon dataset is not something that is native to the OSM dataset, then that makes my point -- there should be one dataset that accurately represents all types of that item, and in this case that would be the points.
My application already has its own ways because they are more accurate and reliable than the OSM ones. Does the OSM community not appreciate that there could be use cases for its data where sometimes someone doesn't need the entire dataset?
I'm happy to discuss this on an email list. I'm guessing you're more familiar with what list that could be so I'll leave it to you to post.
-
Comment from SomeoneElse
Just spotted this discussion and thought that it was perhaps worth mentioning that osmconvert (see http://wiki.openstreetmap.org/wiki/Osmconvert ) can easily extract centroids of OSM ways to make it easy to process. There's no need (in fact in most cases it's actually confusing) to add details both to a way and a separate node in the middle.
-
Comment from Peter Dobratz
I posted on my and there's some discussion there:
http://www.openstreetmap.org/user/Peter%20Dobratz/diary/35170 -
Comment from SK53
See my recent blog post (http://sk53-osm.blogspot.com/2015/05/retail-outlets-on-openstreetmap.html) about creating a worldwide dataset of retail POIs as centroids using osmconvert. This demonstrates that there is absolutely NO NEED to have everything as nodes.
The addition of a duplicate node for the polygon, if repeated worldwide would seriously skew statistics & force complex processing to determine if a node exists in a similarly tagged polygon.
This node should really be deleted.
-
Comment from Paul Cone
It seems like a lot of work to just get a set of point which represent all hardware stores. Maybe there is "NO NEED" to have them in the raw data but there certainly is a need for a simpler process. I will probably just go buy data from InfoUSA rather than continue to argue this with OSM folks.
-
Comment from SK53
If you just want hardware stores use Overpass with the centroid output & you can have them all as JSON, Or even ask me & I can give you a list of hardware stores as centroids (Or you can purchase data from suppliers such as Geofabrik & OpenCageData). OSM is a vast shared database. If you try & make it work for your own personal needs you may break it for others needs. It is emphatically NOT in its raw form likely to be directly suitable for a given data consumer. The idea is that we provide the data & people build services on top of that for needs such as yours.
Nodes (1)
Welcome to OpenStreetMap!
OpenStreetMap is a map of the world, created by people like you and free to use under an open license.
Hosting is supported by Fastly, OSMF corporate members, and other partners.
https://openstreetmap.org/copyright | https://openstreetmap.org |
Copyright OpenStreetMap and contributors, under an open license |