Is EdgeIndexMap needed in r_c_shortest_paths ?

[andrea-cassioli-maersk]

I have been using r_c_shortest_paths for a while, and recently because of a code refactoring I realised that one of the template parameters, namely EdgeIndexMap, seems not to be used anywhere in r_c_shortest_paths. I was then wonder whether I am missing its purpose. Could anyone shed a light?

[jeremy-murphy]

Interesting. I would have to dig into the commit history to find out why.

[andrea-cassioli-maersk]

Hey @jeremy-murphy , any luck finding more about this?

[jeremy-murphy]

Sorry, I haven't had a chance to work on it.

[andrea-cassioli-maersk]

So I realised this is hanging and I have just tried removing the EdgeIndexMap from all the version of rcsp in version 1.86.0 and seems to work. As far as I can see there is no use whatsoever.

In the first version I can find

bb1977b#diff-bf4147b63ecc94907ce57f721126a699ae6f8696efbe83ea87dfdadc3706ab0f

there is no use for the edge map anywhere.

[andreacassioli]

I gave it another look. I quite sure it is not use. In the very first commit the parameter did have a variable name also in the dispatch, but still the edge map was unused.

I guess for sake of compatibility it is not a question of removing it, but maybe a note in the docs?

[jeremy-murphy]

I would prefer to fix it completely, even if it means a breaking change.

[andrea-cassioli-maersk]

I would prefer to fix it completely, even if it means a breaking change.

I will give it a shot. Do you mind a bit of code upgrade to c++14 (that the current minimum standard, right?)

[jeremy-murphy]

Please go ahead.