[Openstreetmap] Unified data-gathering and editing application

SteveC steve at fractalus.com
Sun Feb 27 20:30:56 GMT 2005


* @ 27/02/05 10:56:35 AM nick at hogweed.org wrote:
> Been talking to Steve about the possibility of a common data repository for 
> OpenStreetMap and my Freemap project, /9and any other free mapping projects 
> for that matter...) Different websites and desktop apps could then use 
> whatever data (i.e. street, countryside or whatever else) they need, through 
> the use of SOAP or similar technology.

Hello everyone, I've had a diet of lemsip with a bad cold for almost a
week. Its great if you can live on vitamin c and paracetamol!

Anyway. I think the central data store is pretty central to this whole
idea, something I'll talk about at the opengeodata meeting thing.
Network effect of users and data means that there is no real comparison
between Amazon and whoever is it's no.2, wikipedia and it's no. 2 and so
on.

I think Metcalfes law (IIRC: the utility of a network rises as the
square of its nodes) applies on multiple levels: the number of people
editing the same data, the number of apps able to edit the data, the
number of streets etc entered all get helped by a central store whatever
your philosophy on application type or data collection method.

I think the same idea applies here and why I think openstreetmap would
work well as that store.

openstreetmap.org is hosted at a university that isn't going away, for
free with tons of bandwidth, cpu and database space. As a forum for
discussion its worked pretty well. The next stage is to try and get
things working, or quite a few people already have.

In that context I'd like to reiterate my basic philosophy behind this: I
don't much care for which language, mapserver or application people use
or whether people want to trace old maps or use gpx track tracing or
whatever. I would like to facilitate any of that through an open central
store, to get more free open geodata.

In that vein, I again offer anyone subversion accounts, accounts on the
machine and so on to develop stuff in whatever language you like. Of
course some people don't need that kind of stuff.

In the absence (until recently) of anyone wanting to do stuff like this
I went ahead and implemented a basic xmlrpc api and the applet which
might have labeled me a java nut to be avoided. Hope it hasn't.


So I should ask, how can I do better at this?

I'll take one stab: we need to have a well thought out xmlrpc api. I
don't think there is much antagonism against xmlrpc, as there are lots
of libraries for every language etc.

To date I've just implemented what I needed. It is not hard at all to
add new data types (I'm going to add point of interest etc soon) or
methods and do it with a wiki-like history (which everything has). The
question is what do people want from it?

The question about ontologies also bears thinking about. I'm a much more
bottom up person and it feels a bit like the Open Directory Project
compared to del.icio.us - so why not make the ontology editable and
re-classifyable too? I have some vague thoughts on this but I've written
too much already :-)

> One key component of gathering and processing the data would, I feel, be a 
> unified desktop application which can do several things, namely:
> 
> - read data from a GPS;
> - allow the user to edit the data (waypoints, routes, tracks);
> - allow overlays of Landsat data, NASA SRTM height data, and any other 
> freely-available data for estimation of not   easily-surveyable features e.g. 
> woods, lakes, hill summits on private land;

I think such a thing would be tres tres useful and much faster than the
applet.

have fun,

SteveC steve at fractalus.com http://www.fractalus.com/steve/




More information about the talk mailing list