Hide the traits class parameters from the doc

[Peter Kovacs]

The last template parameter of many algorithm classes (Bfs, BfsVisit, Dfs, DfsVisit, Dijkstra, BellmanFord, Preflow, Circulation, MinCostArborescence) is a traits class. It is always shown in the doc, but it is typically not documented. This could be misleading, since the paramater seems to be mandatory.

A good solution would be to hide all these template parameters entirely.

This ticket is a follow-up of #179.

[Alpar Juttner]

Replying to kpeter:

A good solution would be to hide all these template parameters entirely.

To be precise, this is only a satisfactory solution. The good one would be to fully document these traits class parameters.

[Peter Kovacs]

Replying to alpar:

To be precise, this is only a satisfactory solution. The good one would be to fully document these traits class parameters.

Recall that they were fully documented in the 0.x releases, and we decided to hide these parameters, see the changeset in #185. Just like the basic iterator interfaces of the graph structures: they are needed to implement e.g. adaptors, but we do not document them.

I don't think we should change our mind and revert these parts of the doc.

[Peter Kovacs]

Another solution is to list these template parameters in the doc for all classes, saying something like that: "The traits class that defines various types for the algorithm. This parameter is typically not set directly, consider to use the named template parameters instead."

[Peter Kovacs]

This task should be done after the changesets of #179 and #180 are merged into the main branch.

Which solution do you prefer? Hide these parameters entirely or document them?

[Peter Kovacs]

The attached patch [70e6378230c3] adds a proper documentation to all these traits class parameters.

[Peter Kovacs]

[75e6020b19b1] is a slightly better version, consider to use this one.

[Alpar Juttner]

Replying to kpeter:

[75e6020b19b1] is a slightly better version, consider to use this one.

It is in the main branch.