Franco Fusco

Hi @wxmerkt! > The current implementation does not check the output of angles::shortest_angular_distance_with_large_limits which indicates if the returned distance is an unspecified (and potentially wrong) case. You are totally right,...

Another important point: when I introduced the `large_limits` functions, I updated three controllers (those that were using `shortest_angular_distance_with_limits`): - `effort_controllers/JointGroupPositionController` - `effort_controllers/JointPositionController` - `velocity_controllers/JointPositionController` Once we have decided how to...

> I noticed that for some cases with very, very large limits we had the opposite direction being returned as intended - an error is returned and defined in any...

> I think the confusion here might be that the documentation of shorted_angular_distance_with_large_limits doesn't make it clear what shortest_angle will be when it returns false. From peeking at the code,...

I had a similar problem yesterday, but with limits set to `-20*pi` and `20*pi`. I know they are absurd, but in my opinion the problem had to be addressed (at...

Just a short update about the PR ros/angles#16 I opened at [angles](https://github.com/ros/angles): after the first review, I followed the suggestions and the "fix" for large rotation limits will be provided...

The problem should have been fixed by #466, but I do not know if it is already available in the binaries as well... If not, you can always clone the...

In my case I have a local copy of KDL - I wrote the solver for trees so it was convenient for me to have it locally. As far as...

I manually added the sources for the Tree solver: they are all inside the `src` directory, so that they cannot be included in another package by mistake. In principle, once...

Thanks for all the feedback, I will try to address these points in few more commits!