Doc improvements in Bfs, Dfs, Dijkstra

[Peter Kovacs]

Bug fixes in the function type interfaces

[Peter Kovacs]

Doc improvements (only for the main branch)

[Peter Kovacs]

[a4d00df8852e] contains fixes for the function type interfaces. These bugs are not so critical, but the patch may worth to be merged into the release branch 1.1.

[f85e505e7eba] contains doc improvements (on the top of the former patch) inspired by the doc of the ported BellmanFord class (see #51).

[Alpar Juttner]

Replying to kpeter:

[a4d00df8852e] contains fixes for the function type interfaces. These bugs are not so critical, but the patch may worth to be merged into the release branch 1.1.

I don't understand what kind of bugs are fixed by this chgset.

[Peter Kovacs]

Replying to alpar:

Replying to kpeter:

[a4d00df8852e] contains fixes for the function type interfaces. These bugs are not so critical, but the patch may worth to be merged into the release branch 1.1.

I don't understand what kind of bugs are fixed by this chgset.

I don't know if these mistakes should be called bugs or not, that's why I did not use this word in the commit log.

For the function type interfaces of Bfs, Dfs, Dijkstra and BellmanFord there are several named parameters for setting attributes (with const T& parameters) and there are two named parameter functions for getting results, namely path() and dist(). They also had const T& parameter and const cast was made for them. So the doc of these functions were misleading, moreover a code like that can be compiled:

dijkstra(g,length).path(Path()).run(s,t);

These mistakes (that were actually made by me) were fixed in the changeset. As you can see, the test files should also be modified. The original versions should not be compilable.

[Peter Kovacs]

Replying to kpeter:

Replying to alpar:

Replying to kpeter:

[a4d00df8852e] contains fixes for the function type interfaces. These bugs are not so critical, but the patch may worth to be merged into the release branch 1.1.

I don't understand what kind of bugs are fixed by this chgset.

I don't know if these mistakes should be called bugs or not, that's why I did not use this word in the commit log.

Now I don't think these "mistakes" were mistakes at all.

Currently all the named parameters (predMap(), distMap(), processedMap(), reachedMap(), path(), dist()) have const T& parameters, thus temporary objects can be passed to them (as I wrote in the above comment). In some cases it could be useful for named parameters that work with maps. E.g.

   std::vector<Node> v(countNodes(g));
   dfs(g).processedMap(loggerBoolMap(v.begin())).run(s);

So const T& parameters should be kept for maps. For path() and dist() I don't know such a use case, but it could be imagined. Therefore I reject my own patch (however the doc can be misleading in its current version).

[Peter Kovacs]

Doc improvements (without modifying the interface)

[Peter Kovacs]

I suggest applying [f47b6c94577e] instead of the two former patches: it contains doc improvements, but does not modify the const parameters.

[Alpar Juttner]

Replying to kpeter:

I suggest applying [f47b6c94577e] ...

Done.