On 27/05/2015 09:57, Daniel Koć wrote:
W dniu 27.05.2015 9:38, Jean-Marc Liotier napisał(a):
Also, the address must be unique - it must not be present on more than
one object, even if more than one POI exists at that address.
So there are exceptions to this rule: I know at least one example
where the same address is given for two different retail/service
buildings (~500 m in city along river is rather a big distance, and
it's not a tagging mistake).
So now I understand why some information systems I know use for the
address a primary key distinct from any of the address's components.
Using a random primary key also makes the address still relatable across
events such as street name change.
On 27/05/2015 09:48, Martin Koppenhoefer wrote:
Am 27.05.2015 um 09:38 schrieb Jean-Marc Liotier <j...@liotier.org>:
Also, the address must be unique
why?
Relational integrity ? But after Daniel's remark above, I now believe I
was mistaken in how I understood address uniqueness: the address object
that must be unique but two addresses may have the same attributes.
When French telcos send each other messages to order construction of
local loops, they mention an address identifier - the address attributes
are accessory.
_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
https://lists.openstreetmap.org/listinfo/tagging
