[OSM-talk] keys

[Immanuel Scholz]

Hi,


> 1. No structure to keys or values - anything goes aka flickr - and
> therefore not much different than where we are now.
> 2. A registry approach that provides additional information for keys (and
> values?) to be interpreted - (Is that the correct interpretation? I could
> use some more info on how this works in practice - or was it documented
> somewhere?).
> 3. Expanding keys and values, each key and value potentially containing
> more than one piece of information... sounds ugly.
> 5. Something else even more complicated.

Where is 4?  ;-)


Well, anyway. 1 and 2 could be the same.

Or other to say: You cannot forbid people to install a registry holding an
ontology somewhere and use this together with the existing data. In fact,
such a thing is already deployed in three places: Map Features, the
osmarender-config file and the gui-descriptors for the online applet
(defining stuff like "name", "class" and "oneway" in the basic tab).



What Etienne complains about is the lack of specification whether a set of
key/values fit to a given ontology. So as example it is currently not easy
for osmarender to say, whether object xy was tagged with the
osmarender-compatible scheme in mind.


> So I want it to remain a simple method that is easy to understand and use
> but clearly there is a benefit in improving the structure or the
> validation of at least the keys.

My plans for JOSM were about the following:
I want to give advanced users the posibility to define a form where users
can enter the keys/values. As example, by configuration (don't know how to
do, maybe html-forms, xml or Java-classes...) you could create a form like
this:

Name:  [...........]
Class: [minor...][V]
[ ] Oneway
[ ] roundabout


much like the current applet's basic tab. Users can redefine, load/save
presets of this at will..
These may also be coupled with preference settings like colors.

Doing this, I want to simplify the key/values for beginners without adding
ontology complexity to JOSM, the data format or the data structure itself.

Ciao, Imi

PS: Another thing to remind: There WILL be some thing of differnt
namespaces sometimes. This will be the time when the second osm - server
fork has been opened featuring some very crude scheming (did someone said
"slopes"? ;). Then the "namespace" is the server, the annotation is from.