[OSM-talk-be] Fietsknooppunten, bicycle nodes "Roundabouts and around squares"

marc at vmarc.be marc at vmarc.be
Sun May 17 13:19:04 UTC 2015


Jo,

> What do you use at the moment?

The osma.vmarc.be analysis runs unattended in a virtual server 
(http://www.versio.be/cloudbox). Every night at 01:00h a job is started 
to load fresh data from Overpass API (http://overpass-api.de/). The data 
is analyzed using logic written in the Scala programming language 
(http://www.scala-lang.org/). The Scala analysis result objects are 
transformed to json format (http://json.org/) using Spray 
(https://github.com/spray/spray-json) and stored in a Couchdb database 
(http://couchdb.apache.org/).

The web interface is using the Play framework 
(https://www.playframework.com/) for Scala.

At this moment I am installing a full history local copy of the Overpass 
API database (with minutely diffs). This should allow for more frequent 
and more in-depth analysis without increasing the load on the public 
Overpass API instance.

> I'd be inclined to do it all within JOSM. Unfortunately that makes 
> running it as a 'batch job' automatically harder.

To have the analysis running without needing manual user intervention is 
essential I think.

> If I would perform the analysis, in what format do you need the 
> results?

The analysis and presentation code are cleanly separated but do share a 
common set of Scala domain objects that match pretty much what you see 
in the different pages on the web-site. I guess in theory the json 
format as stored in couchdb could serve as an interface point for 
analysis results that are generated in a different way, but I am not 
sure if this would be better than publishing those results separately.

Kind regards,
Marc




More information about the Talk-be mailing list