[Openstreetmap] openstreetmap.org updates, ontology stuff

[Jo Walsh]

hello,

On Sat, Mar 19, 2005 at 10:18:25AM +0000, Nick Whitelegg wrote:
> On Saturday 19 Mar 2005 00:34, Matt Amos wrote:

> That's what I do with my Freemap data ATM and I think it's what Steve is 
> intending to do.

the bitwise stuff scares me. i really like things to be
human-readable... ;) i use URIs for type indicators and identifiers.
(though i realise there might be a performance drag.)

http://uo.space.frot.org/freemap/tracks.sql
is the schema i am using with postGIS to store this information.
i am very happy with the spatial query capacity in postGIS and would
hesitate to step back from it. its functions few of which i have
really used: http://postgis.refractions.net/docs/ch06.html#id3363858

I found it quite simple to turn mapserver into a WFS client or server.
http://mappinghacks.com/projects/mapscript/03_mapserver_wfs.html /
http://uo.space.frot.org/freemap/map.cgi?service=WFS&request=GetCapabilities

I was able to load gps tracks over WFS from one machine and draw over
them and save the new dataset on a different machine. 

http://geoserver.sf.net/ is a fairly lightweight java (cough) servlet
that can sit on top of several different kinds of spatial data stores
including Arc* and now has support for MySQL

WFS is good because you can load it on the fly or cache it /
aggregate it, but the process for doing the latter is not fully spelt
out yet :/ also GML has its limits and is not fully supported by a lot
of tools. 

We don't need a national spatial data infrastructure but we have a lot
of options for sharing data and it would be nice to pick ones that are
sustainable:

1/ Database sharing, via sql dumps or direct access, with common
schema
2/ Track or waypoint-segment sharing in XML over HTTP+something
3/ Exposing data over WFS and caching it appropriately.
4/ Publishing ESRI shapefiles over HTTP

This is influenced by whether you're at the track-gathering,
line-segment-inference stage of a process or a
road-segment-annotating, routing-application-writing stage. 

There's no reason these two 'stages' can't be nearly simultaneous or
use the same data transmisison mechanisms.

I don't have a huge amount in the way of tracks, personally. I
gathered some nice tracks in bristol and of the motorway and the train
tracks going back and forth from London. 

I uploaded some random tracklogs that i found on my hard drive to
http://uo.space.frot.org/freemap/tracks/

They are not indexed, it would probably be easy to generate a little
script that showed a bounding box for each one on a little map.

For tracks that get uploaded via a cheesy web UI, they are stored in
the schema shown above. I could save the original gpx file to a public
index, too. This isn't running yet, i'm hoping to get the bristol
wireless people to test it out, and write a user-login hook around the
upload form.


-jo