Andrew, lets made an exercise. http://www.openstreetmap.org/way/200522987 Can you tell me all addresses for italian restaurant (http://www.openstreetmap.org/node/2475029870)? (Let's assume that we have 3 additional address points)
and the same story for "Сосед" store here http://www.openstreetmap.org/way/90681689 ---- For the first example we'll get three addresses: Tallinn Pikk 1 Tallinn Nunne 2 Tallinn Lai 2 And, I think it would be correct. As for second example we will get (I have translated street and city) Kaliningrad Lenin's ave. 83a Kaliningrad Lenin's ave. 83b Kaliningrad Lenin's ave. 83v Kaliningrad Lenin's ave. 83g Kaliningrad Lenin's ave. 83d And this is actually wrong. ---- And we can't choose correct algorithm, or it will be full of code like this: if (we_are_in_Kalinigrad_and_they_usually_use_next_approach) { //threat points enclosed by outline in one way } else if (we_are_in_Tallin) { //threat points enclosed by outline in other way } else if (we_are_in_Austria) { //threat conscription numbers in one way } else if (we_are_in_Czech_Republic) { //threat conscription numbers in other way } With addrN you are able to add addresses explicitly to every poi, as good as for building, no matter how many addresses or how complex addresses scheme you have. Also you could specify interconnections between addresses rows. And solve controversies between points inside outline and/or addr on outline it self. This example is not the only one. So, again, nodes are not enough. And Andrew, I have provided rather detailed examples, so the answers like "nodes are enough for anything and everyone" or "it's just a conner case, just forget about them" without any further details will be just unfair. Mon, 19 Jan 2015 13:02:39 +0100 от Andrew Shadura <and...@shadura.me>: >Dmitry, >In most cases the address of a POI is the same as the address of a building >it's located in. The address of a building can be derived from address nodes >within its outline ', and then extrapolated to other POIs in the same >buildings. It doesn't work precisely when there's no outline, but in this >case you've already lost the precision. >Moreover, this operation has to be performed anyway to cope with really bad >and incomplete data, so there's no reason to not re-utilise this functionally >once again. >-- >Cheers, > Andrew
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging
