[OSM-talk] Map Features tagging question

[Andy Robinson]

Nicola Ranaldo wrote:
>Sent: 20 July 2006 19:25
>To: talk at openstreetmap.org
>Subject: Re: [OSM-talk] Map Features tagging question
>
>On Thursday 20 July 2006 10:28, Andy Robinson wrote:
>> No, you are absolutely right, nearly everyone wishes to see a level of
>> consistency. The trouble is not everyone wants to use the OSM data for
>the
>
>[...]
>
>> tags that are already there. Existing tags should never be deleted if
>> possible in case they are being used by someone.
>
>From a developer point of view:
>
>* when a user enter a tag for an osm object, this information is available
>for
>everyone in the world
>* this information should be useful for the community and not for private
>needs (use your hard disk for that :))
>* this information in the 99% of cases will be used by the software and not
>by
>humans!
>* the software does not know the human meaning of a tag but has to follow a
>well-defined set of rules
>
>==>
>
>* you should enter data in osm if well defined!
>* you should fix data in osm if not well defined!
>* before adding an undocumented tag in the system post an rfc on the
>mailing-list
>
>If this is against the freedom (to construct an unusable data set?), we
>should
>separate tags in "namespaces", and reserve an XXX official namespace only
>for
>common rendering/viewing/routeplanning where data integrity is strongly
>enforced. Other namespaces could be free (to construct private data set?)
>if
>we really need them. Editors and viewers could be free or XXX compliant.
>
>The same concept should be applied to data primitives, we should strongly
>define them and enforce data integrity on the database, for example the id
>is
>an integer > 0 and in the osmplanet there are ways with segs with a
>negative
>id. Are two segments from x to y and from y to x admitted? Are two nodes
>with
>the same lat,lon admitted? Actual editors take care of those?
>As specified in the "Data Primitives" wiki page, ways and areas are objects
>with the same elements. Is necessary to separate them? If not what are the
>differences? these are simple examples, but while coding osm "clients" they
>affects data structures more then a bit, and shows strange paths when doing
>route-planning!
>
>	Niko

All of this is restricting what the data structure beyond the nodes,
segments, ways and areas should be and that's not what I'm comfortable with.
I'm very happy with the flexibility to tag as I wish for whatever purpose I
wish and that was always the intention of OSM. OSM started out on the basis
that it would not impose any structure on the tagging. It's up to users of
the data to do that if they wish it to have such structures. So far nobody
that I am aware of has set up a rigid structure for that specific purpose,
but there is nothing stopping anyone from doing it, so long as they don't
mess with the tagging that's already there. Personally I like a bit of
formality to get the ball rolling and then plenty of flexibility to do
whatever I like after that but I know that if I was writing navigation
software I would need to tag the data specifically for that purpose.

Map Features was set up to focus upon rendered maps. We need more to get
interested in establishing other tagging rules for other purposes as I don't
think you can bank on OSM ever doing it for you. If it subsequently emerges
that all the map Features tagging needs to have an MF: namespace or
something to avoid too much confusion then I have no problem with that but
please don't shackle the data beyond that.

>
>
>_______________________________________________
>talk mailing list
>talk at openstreetmap.org
>http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk


Cheers

Andy

Andy Robinson
Andy_J_Robinson at blueyonder.co.uk