Named Parameters
| Named parameters are a legacy API. New code should use the positional overloads documented on each algorithm page. |
Many of the Boost.Graph algorithms have a long list of parameters, most of which have default values. This causes several problems. First, C++ does not provide a mechanism for handling default parameters of template functions. However, this can be overcome by creating multiply version of an algorithm with different numbers of parameters with each version providing defaults for some subset of the parameters. This is the approach used in previous versions of Boost.Graph. This solution is still unsatisfactory for several reasons:
-
The defaults for parameters can only been used in a particular order. If the ordering of the defaults does not fit the users situation he or she has to resort to providing all the parameters.
-
Since the list of parameters is long, it is easy to forget the ordering.
A better solution is provided by bgl_named_params. This class
allows users to provide parameters is any order, and matches arguments
to parameters based on parameter names.
The following code shows an example of calling
bellman_ford_shortest_paths using the named parameter
technique. Each of the arguments is passed to a function whose name
indicates which parameter the argument is for. Each of the named
parameters is separated by a period, not a comma.
bool r = boost::bellman_ford_shortest_paths(g, int(N),
boost::weight_map(weight).
distance_map(&distance[0]).
predecessor_map(&parent[0]));The order in which the arguments are provided does not matter as long as
they are matched with the correct parameter function. Here is an call to
bellman_ford_shortest_paths that is equivalent to the one
above.
bool r = boost::bellman_ford_shortest_paths(g, int(N),
boost::predecessor_map(&parent[0]).
distance_map(&distance[0]).
weight_map(weight));Typically the user never needs to deal with the
bgl_named_params class directly, since there are functions
like boost::weight_map that create an instance of
bgl_named_params.