Even Rouault wrote: > Last but not least, is there really an interest for such a driver given all > the existing advanced tools that do similar things : osm2pgsql, osmosis, > etc... ?
Osm2pgsql outputs only to PostGIS and even PostGIS is super fine I consider that the threshold in installing and managing it is too high for many users. In addition, there has been only two Windows builds of osm2pgsql available ever and the latter one is more than two years old. ImpOsm is impossible to make work on Windows due to something called Tokyo Cabinet. Osm2pgsql fails every time for me with my lean Linux box with only 700 MB of memory and I must do local import with my home computer and use PostGIS dump/restore after that. And on my home Windows laptop running import with osm2pgsql (with that 2 years old version) takes something like 3 hours while Sandro had a try with the same Finnish OSM excerpt and reported that he could parse it with ReadOSM in five minutes. If something is possibly 30 times faster than something else I usually get interested. If you count together the number of users who are on Windows OR do not want to learn to be DB admins OR are willing to run cheap and weak virtual servers, then I would say that there might be enough interest for such a driver. I think that what Osm2pgsql is doing is enough for the majority of users. That is, create point, line and polygon layers from features which have at least on tag in the configuration file. Parsing relations must be pretty difficult, at least with polygons and therefore I was thinking that by using the ReadOSM library the responsibility about doing it right could be left to ReadOSM developer. Advanced users could still use the advanced tools. -Jukka Rahkonen- _______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/gdal-dev
