[OSM-talk] Proposed feature for noname

Frederik Ramm frederik at remote.org
Mon Sep 22 01:41:10 BST 2008


Hi,

Gervase Markham wrote:
> Names are unusual in this regard because there are certain classes of
> things which one expects to have them (such as roads)

And that's what I wanted to say with my "tests" idea. It is not natural 
that a road has a name, it is just something that we expect. This 
expectation is in our heads, it's not in the database. We can write a 
test that checks whether the expectation is met - this is what the 
no-name map does.

If you put "this road has no name" in the database, then indeed, from 
the programmer's perspective, why not also "this railroad track has no 
name"?

> So having a "noname=yes" (or whatever - the syntax is unimportant for
> this discussion) absolutely does not mean you need a "nopostbox=yes" or
> a "notractorlane=yes" tag.

I agree that the example was a bit extreme. But still; the concept that 
the absence of names for certain objects is noteworthy while the absence 
of names for other objects, or the absence of other properties for the 
same objects, is not, is a rule (or formalized expectation) you're 
putting up, and it will not be the first such rule.

(I don't care too much whether you follow my suggestion of tagging a 
list of tests to be ignored for an object, or whether you stick with 
noname=yes, because any validator could trivially be made to understand 
both. I just wanted to get across the conceptual message that the idea 
of roads having names is nothing unqiue, just one in a long list of 
things that should go together. Railway stations will usually be 
connected to railway lines but maybe sometimes they aren't; will this 
necessitate a "not_connected_to_railway_line=yes" tag, then? You're 
unlikely to find a protestant chapel in a catholic graveyard and I could 
imagine a validator flaging this as a potential error; if it happens 
nonetheless, should we tag it as "different_denomination=yes"? And so on.)

Bye
Frederik

-- 
Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"




More information about the talk mailing list