CHAPTER 2. VECTOR PRODUCT FORMAT
2.5.2 Feature and primitive table relationships
Afeature id is linked to a primitive id.
Untiled When the coverage isn’t tiled, the relation between a feature table and a primitive
table is quite simple. Each feature entry will have a direct link to a primitive id. This gives
good performance going from feature to primitive, but very poor performance going from
primitive to feature. So when selecting a line, it will take a while before the programme can
determine the feature.
Tiled The link to the primitive is also direct here, but one has to keep in mind what tile
the primitive is in. There can be more than one edg with id “1” since there are multiple tiles
and each tile can have an edge. Like untiled relationship, going from primitive to feature,
this will also have poor performance.
Thematic The thematic index will improve the performance when going from feature to
primitive. For example, a thematic index on the TILE
ID column in the feature table would
provide a list of all records for a given TILE
ID value. If the TILE
ID’s are given, getting all
the features within a given tile will go faster. Not only TILE
ID’s can be used that way, also
Tiled with feature id pointers in primitive tables When one gives the primitive a
reference to the feature, it will improve the performance when a feature has to be determined.
The tile where the primitive is in, is known. That way the tile id can be found in the feature
Tiled with feature id pointers in primitive tables and thematic indexes This is
combining previous with thematic indexes.
This implies that a single feature is composed of many primitives. The structure is about the
same as from one-to-one relationships, but now there is a feature join table.
Many features can relate to many primitives and vice versa. There is also the need for a
feature join table here.