In light of the upcoming virtual meetup on osm2pgsql I decided to write a short blogpost about the top five features I want to see in osm2pgsql.
- To be able to move my map localisation stuff from PostgreSQL stored procedures to the data import stage I will need the get_bbox() feature of the new flex output driver to also work for relations. The reason is, that I do geolocation aware transcription which would not be possible on relations without such a function. This will then hopefully recover my development of the German map style.
- The flex backend has a new feature which will allow to pass information from a relation to its members. Currently this works for ways only but would be very nice to have for nodes and relations as well. What I think about where such a feature will be handy is for rendering of site relations.
- There are many ways to import OSM data into PostgreSQL, but if the target database needs to be kept up to date with osm there are only two options left. These options are imposm and osm2pgsql.Unfortunately there are features which are not available on both applications. Currently imposm will need rougly half the storage size osm2pgsql uses for updating. The main reason for this is the usage of LevelDB which might also be an option for osm2pgsql.
- Another feature which imposm provides is support for generalized tables for rendering low zoom objects which would be nice to have in osm2pgsql as well.
- Finally I would like to see support for geometries to be stored in TWKB format which will help to further reduce the storage size of production databases.