2010/11/17 Ulf Lamping <ulf.lamp...@googlemail.com>:
> You didn't answer my question.
>
> How do you handle "conflicting" grass and trees using landcover?


IMHO they are not conflicting. If you tag grass there is also soil,
there is also microorganisms, etc., and if you're tagging trees there
is surely something at their bottom.

If you wanted to tag these details, (whether there is grass or moss,
or whatever) you will need a different tag/subtag (maybe surface is
not too bad for this, or you use a tag (de:Bodenpflanzen don't know
this in English).


>>>> Natural is IMHO an ideal example of a tag to diffuse
>>>> clarity and create confusion, because it is a mix of all sorts of
>>>> features.
>>>
>>> It contains features that naturally appear. I am not confused.
>>
>> fine, I got this, you don't care for semantics or content of tags, of
>> what the do express, if they are grouped with a certain sense or not.
>
> They *are* grouped in a specific sense. Already. Today. It's just that you
> don't like that group because it doesn't fit into your mental model how it
> should be.


no, if they were grouped in a specific sense, all features that are
"natural" were there, and all features that aren't natural would be
there. It is not like this.


> It is accepting that semantically different things can reside under the same
> key and that this doesn't cause any problems - except for people like you
> that seem to think that a systematic approach is a value in itself.


it depends what you implicate with "creating problems". If you invent
a new value, it is not clear where to put it, because there is no
logic. What about valleys. Say you wanted to put a valley in the db.
Is this natural? Maybe yes, as long as it is not artificial. If you
make presets for an Editor, you will have to filter the values by what
they express, some are geographic features, others are physical
landcover features, others are even different things. This complicates
the life of all: mappers, especially new ones as well as application
developers.



> It is also knowing that remembering six different "semantically correct
> groups" each with their own key are a lot harder to remember than two keys
> containing easy to remember groups of values.


this worked fine in the beginning, with a very reduced set of
k/v-pairs but also very low detail as a result. People didn't take
tags too seriously, e.g. a lawn in a German city could easily become
mapped as village_green or any art-related feaure became an
arts_centre.). The more details we add and the more tags we create,
the more important it becomes to get towards a system that is
inherently logical and consistent. The alternative would be not to
have 2 categories with everything in, but only 1 and leave it to a
tagging-subsystem to create semantics, meaning and order (in the
editors, in the renderers, in the descriptions in the various
languages, etc.).


> Fine. But people wanting to change existing things should have and being
> able to explain good reasons.


I tried to explain the reasons, and sorting stuff in a semantic way is
a good reason IMHO. If you don't see it like this, I'm sorry but I'll
have to accept.

Cheers,
Martin

_______________________________________________
Tagging mailing list
Tagging@openstreetmap.org
http://lists.openstreetmap.org/listinfo/tagging

Reply via email to