[OSM-talk] Advanced Relationships

Frederik Ramm frederik at remote.org
Thu Jun 21 01:31:36 BST 2007


Hi,

>> There are 120 instances of street number data in the whole planet file.
> 
> One effect of non-standardised tags. This bounding box contains nearly
> 600 segments with housenumber data:
> http://www.openstreetmap.org/index.html?lat=53.00124355770805&lon=6.5719647173087985&zoom=11 
> 
> (look for housenr). Who knows where else.

Darn, I only counted "place_numbers", "house_numbers", "first_number", 
"last_number", "number", "left_housenumber_from", "left_housenumber_to", 
and the plain "number".

> I disagree. For a routing application the segment is obviously the
> closest point it's going to be able to route to, even if it is a
> kilometre long.

This is just an assumption you make. I could equally well say that a 
node is "obviously" the closest point a routing application is going to 
be able to route to. And on the other hand I could also use tags on a 
way to specify where something is (like "house number 123 is a hundred 
metres into this way" or so).

> I never said they were. I only said that we need to think of (and
> implement) the alternatives *before* removing segments, not after.

I'm not so sure. It wasn't "we" who suggested that house numbers be 
tagged the way they are, it was some people putting them in. Why should 
"we" now be under an obligation to provide alternatives? The same people 
that had the idea of tagging segments with house numbers will come up 
with something else (probably 20 different ways of doing it). Some will 
use nodes, some will make short ways, some will make long ways and 
invent special tags.

If house numbers were halfway standardised in our current database we 
could think about some way for automatic conversion into something else 
but I don't see that.

I don't want to start a fight over this, as the abolition of segments is 
not something I plan to do tomorrow or so. It's just my principal 
thinking that any future changes in the data model *might* make some 
things impossible that are possible now, and while we will always try 
not to throw away valuable information, there is no *automatic* 
obligation to always support everything that was possible at some time.

(For example, it has just now become impossible to create ways with no 
segments in them. What if someone devised a cool schema to store 
interesting information in OSM using empty ways as a vehicle? He's 
screwed. And nobody has thought of or even implemented alternatives for 
him!)

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00.09' E008°23.33'




More information about the talk mailing list