Version 5.0 The LEDA User Manual

• Classes GRAPH and UGRAPH (sects. 11.2 and 11.5) are templates with two
arguments, vtype and etype, which are reserved for a node and an edge attribute,
respectively. To attach several attributes to nodes and edges, vtype and etype must
be instantiated by structs whose members are the attributes.
• A node array (sect. 11.8) or node map (sect. Node Maps) represents a node attribute,
and analogously, edge arrays (sect. Edge Arrays) and edge maps (sect. 11.12), represent
edge attributes. Several attributes can be attached to nodes and edges by
instantiating several arrays or maps.
Data accessors provide a uniform interface to access attributes, and the concrete organization
of the attributes is hidden behind this interface. Hence, if an implementation
of an algorithm does not access attributes directly, but solely in terms of data accessors,
it may be applied to any organization of the attributes (in contrast, the algorithms in
sect. Graph Algorithms require an organization of all attributes as node and edge arrays).
Every data accessor class DA comes with a function template get:
T get(DA da, Iter it);
This function returns the value of the attribute managed by the data accessor da for the
node or edge marked by the iterator it. Moreover, most data accessor classes also come
with a function template set:
void set(DA da, Iter it, T value);
This function overwrites the value of the attribute managed by the data accessor da for
the node or edge marked by the iterator it by value.
The data accessor classes that do not provide a function template set realize attributes in
such a way that a function set does not make sense or is even impossible. The constant
accessor in sect. 13.14 is a concrete example: it realizes an attribute that is constant over
the whole attributed set and over the whole time of the program. Hence, it does not make
sense to provide a function set. Moreover, since the constant accessor class organizes its
attribute in a non-materialized fashion, an overwriting function set is even impossible.
Example: The following trivial algorithm may serve as an example to demonstrate the
usage of data accessors and their interplay with various iterator types. The first, nested
loop accesses all edges once. More specifically, the outer loop iterates over all nodes of
the graph, and the inner loop iterates over all edges leaving the current node of the outer
loop. Hence, for each edge, the value of the attribute managed by the data accessor da is
overwritten by t. In the second loop, a linear edge iterator is used to check whether the
first loop has set all values correctly.