Inconsistency in Figure 6. GeoPackage Minimal Tables for Tiles

[xeonx]

In my understanding as the primary key of gpkg_tile_matrix_set is the same as the one of gpkg_contents (table_name) the association in the diagram should be 1..1 instead of 1..0-*

An other possibility is to have srs_id column included in the gpkg_tile_matrix_set primary key, allowing to store multiple projections of the same content. But in this case the srs_id column shall also be added in the gpkg_tile_matrix table and primary key.

[jyutzler]

I talked this over with @pdaisey today and we determined that the associations in question should be 1..0-1. The reason for the '0-' is because a gpkg_contents entry doesn't necessarily correspond to a gpkg_tile_matrix_set entry (it might be features instead). We suspect that the '*' multiplicity is an oversight that was not corrected when we prohibited multiple geometry columns per feature table.

It will take me longer to acquire and install the software used to make this diagram than it will to make the actual changes.

[jyutzler]

See the Model branch.
https://github.com/jyutzler/geopackage/blob/model/images/geopackage-features.png
https://github.com/jyutzler/geopackage/blob/model/images/geopackage-tiles.png
https://github.com/jyutzler/geopackage/blob/model/images/geopackage-uml.png

[jyutzler]

Approved in SWG today. See PR #126