[OSM-talk] Users, contributors and developers

[Andy Robinson]

Frederik Ramm wrote:
>Sent: 27 April 2007 11:19 AM
>To: Sebastian Spaeth
>Cc: talk at openstreetmap.org
>Subject: Re: [OSM-talk] Users, contributors and developers
>
>Hi,
>
>I'll try to summarize the namespace/subkey argument for the benefit
>of the "laymen" and using Spaetz's examples:
>
>> 1) Bilangual place names have been mentioned:
>> place=town,name:en=blah,name:welsh=blubb,name_default=welsh
>
>This, for example, would not be possible if you decree the third
>column to be a "namespace". By defintion, a namespace puts the *key*
>into a context - meaning that a key in namespace A may have a
>completely different meaning from the same key in namespace B.
>
>If you intend to separate e.g. English names from Welsh names (or
>metric units from imperial units), then a namespace doesn't help you,
>as you want to put the *value* into a context ("the name if this
>place is: in English, soandso; in Welsh, soandso; in ..."). This is
>what people often call a "subkey", a logical area below, not above
>the key. The major property of the road is that is has a name; this
>name comes in different flavours.
>
>> 2) Renderer specific hints:
>> renderName:osmarender=no
>
>A clear case for namespaces (as is Nick Whiteleggs's "pathtype" tag
>in the "newforest" namespace, because one can easily imagine other
>namespaces where "pathtype" has wholly different meaning and set of
>allowable values than in the New Forest).
>
>> 3) Introducing multiple restrictions on one road:
>> restriction:1=noaccess,restriction_valid:1=6am-6pm
>> restriction:2=oneway,restriction_valid:2=6pm-6am
>
>This has nothing to do with namespaces. Neither "restriction" nor "1"
>and "2" in your example would make sensible namespaces. What you want
>here is a grouping of keys *per object*:
>
>highway=residential
>name=Brauer Blvd.
>(restriction=noaccess restriction_valid=6am-6pm)
>(restriction=oneway restriction_valid=6pm-6am)
>
>Not a good example, as the oneway restriction proably doesn't vanish
>as long as the noaccess restriction is in place, but I know what you
>mean.
>
>> How it's done in the backend and what's it
>> called like should be of no concern to the end user...
>
>What you seem want is a joker column that can take any number of
>forms and uses depending on the context ;-)
>
>
>But I was just trying to summarise the argument. I am actually (by
>now) quite painless about it; we can just introduce a third column
>and see where we get. I suspect we will sooner or later have a
>conflict between the subkey and namespace worlds (i.e. someone
>creating a namespace for the new forest then finding out he needs
>names in different languages), and then we'll have either
>column1=newforest column2=name:en column3=something or
>column1=newforest:en column2=name column3=something or
>column1=newforest collumn2=name column3=en:something, and somehow it
>will sort itself out.
>

Am I missing something here, if we allow the key column to take any format
of tags separated by colons and the value column separated by semicolons do
we not get all the flexibility we need without introducing unnecessary
restriction?

Cheers

Andy

Andy Robinson
Andy_J_Robinson at blueyonder.co.uk


>Bye
>Frederik
>
>--
>Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00.09' E008°23.33'
>
>
>
>_______________________________________________
>talk mailing list
>talk at openstreetmap.org
>http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk