[OSM-dev] OWL + OSM Activity Server

Paweł Paprota ppawel at fastmail.fm
Fri Oct 19 12:15:47 BST 2012


On 10/19/2012 04:02 AM, Alex Barth wrote:
>
> On Oct 8, 2012, at 4:00 PM, Paweł Paprota <ppawel at fastmail.fm>
> wrote:
>
>> On today's EWG meeting we briefly discussed the "OWL-powered
>> activity/history tab" Top Ten Task.
>
> I want to loop back to the history tab on openstreetmap.org and the
> ability to only show changesets that affect data in a bounding box
> that Andy and Pawel brought up in [1].
>
> This specific functionality seems like something we'd want directly
> supported in the railsport, no? Or is it so expensive that it has to
> be tiered out into its separate application?
>
> [1]
> http://lists.openstreetmap.org/pipermail/dev/2012-October/025948.html

I agree with Tom and Roland that this functionality belongs to a 
third-party tool - the apidb schema is not really suitable for 
implementing this kind of thing.

What I think would be good to build:

1. Database schema geared towards analyzing changes in OSM data - that 
is a schema where you can easily fast forward through what was going on 
in OSM and get some additional "context" information for each change.

Ideally of course you would be able to "checkout" a given revision of 
the whole database for any given point in time but that seems like a 
monumental task. Right now OSM data is versioned like CVS - each object 
separately, there is no notion of a "database revision" like you get in 
SVN or Git (think tree for given commit).

This maybe comes back to the discussion of changesets not being atomic. 
Which in itself isn't so bad (other than a misleading name...) but the 
really tragic thing about OSM changes right now is that there is no easy 
way to sort through atomic changes (single changeset uploads) - they get 
no id so I guess you're left with trying to figure it out by 
timestamp+userid combination?

2. A simple way to instantiate and populate such database - just as you 
do with apidb/PostGIS schemas now using e.g. Osmosis.

So in the first version of such schema+tool it would be possible to get 
at least some additional information about elements that are in a 
specific change, for example when there is a modification of node X it 
would be possible to get ways and relations for that node in that point 
in time.

I would gladly get involved in such project since it would allow me to 
offload a major portion of my work with "changeset activities" to this 
new tool/schema/server/whatever we call it.

I think it could be based on OWL however I'm a bit put off by the C++ 
part to be honest - I don't think the XML parsing and database 
operations part really needs to be in C++ since it is certainly not a 
bottleneck. I will try to speak with Matt and find out if he would 
accept changing the C++ part into Osmosis plugin (which is something he 
mentioned to me I think on IRC but said there were no hooks in Osmosis).

Paweł



More information about the dev mailing list