>> >> Technically there are already some shapefile-based datasets in mapnik >> renderer (coastlines hosted in SVN); what is missing is a scalable >> interface to add other external datasets. > > There are a number of organisations attempting the same already. For example, > the "WeoGeo marketplace" hosts many free shapefiles, and so does a github > repository run by Development Seed/Mapbox (who want to make it easy for > TileMill users to access shapefiles). Even ESRI is in that game.
Yes, this is true. It would be great if some of these could be used. What I'm afraid is that many (if not all) of them have conflicting business plan, they are not really interested on sharing their datasets to OSM renderer. >> There are several opensource GIS server packages which could be good >> fit, Geoserver seems to be closest to me. Sure, it has a lot of GIS >> stuff we don't really need right away, and may miss some features, >> but it seems to have good base to get started with it. > > But don't kid yourself - the amount of free datasets around is huge and will > easily use up any capacity we can build up. I would of course much prefer > such a solution to everybody dumping their stuff into OSM because then it is > easier to access (many people behind imports to OSM don't even want to bother > setting up their own renderer - they believe that once stuff is in OSM, we'll > conventiently render it for them). But even this approach has its limits, and > we're likely to reach them quickly. The basic question remains the same: Is > it part of our mission to collect and re-distribute free data that we're not > the owner of? One important part of our mission is to server the community. What community wants is to make nice map, so they create map for the renderer, and so they import for the renderer. And by renderer they don't mean own custom one. Currently only way to improve the map is real import with all its problems, but if Mapnik main renderer could read directly community-maintained directory of "external shapefiles" (whatever the best technology would be) then we would solve the map improvement urge for many, maybe most of the current importers. I don't think that raw disk capacity is really that big issue, disk is quite cheap. Bigger problems are organizational ones like who and how will maintain the datasets, as they need reviews and approvals for Mapnik layout, you will easily have overlapping datasets for same features (and overlappings with OSM manually created/imported data), need to check license compatibility etc. /Jaak _______________________________________________ Imports mailing list [email protected] http://lists.openstreetmap.org/listinfo/imports
