-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Andy Allan schrieb: | Jon, | | Given that #relations < #ways < #nodes, and you (generally) need to | know about all the relations before you process the ways, and you need | to know which ways you are interested in before deciding which nodes | to keep - would it not make sense to dump the relations into the file | first, followed by the ways, then the nodes last? | | Albeit I'm only thinking of osm2pgsql, maybe there's other things that | wouldn't work this way round. | The obvious solution to me would be to abandon the monolithic planet file altogether or at least provide separate files for nodes and ways, too. programs optimized for planet file consumption (in the known order) would then keep using the monolithic file, while other applications could freely use the separate files.
.oO(The only thing I can see with using the "nodes first" approach is, that it might easier to cut bboxes from the file, since the nodes are the objects carrying the georaphical positions. IIRC it is possible to simply concatenate bzip2 files to get one large file that will, upon decompression, result in the uncompressed files concatenated. Then you only have to deal with stuff like where to put the <osm> tags in the separate files. (or you would need to filter them on applications consuming this sort of data)...) - -- Dirk-Lüder "Deelkar" Kreie Bremen - 53.0952°N 8.8652°E -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.7 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFIAl+wFUbODdpRVDwRAjjhAJ0T0JWmlw5uPsO+XZHwsSffEYL64ACgwHTG 4p6F9K5gW/57rpJQuQ8jeE4= =N6Fh -----END PGP SIGNATURE----- _______________________________________________ dev mailing list [email protected] http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/dev
