Catch writes to 'unwritable' addresses on the ROS side and print informative error

[gavanderhoorn]

With #355 basic read/write of individual IOs was added to motoman_driver.

As mentioned in the .srv, writing is only supported to certain ranges of addresses:

https://github.com/ros-industrial/motoman/blob/591a09c5cb95378aafd02e77e45514cfac3a009d/motoman_msgs/srv/WriteSingleIO.srv#L9-L12

Instead of letting writes requests to other addresses through, the driver should perhaps check which address is targeted and implement a fail-early policy (don't even forward the request to MotoROS).

It could then also print a better error message than just the current:

https://github.com/ros-industrial/motoman/blob/591a09c5cb95378aafd02e77e45514cfac3a009d/motoman_driver/src/joint_trajectory_streamer.cpp#L600

a more informative error would include why the write failed.

[gavanderhoorn]

@ted-miller @EricMarcil: would you happen to know the ranges which are writable, or should we look it up in the manual(s)?

[gavanderhoorn]

Related ROS Answers Q&A: motoman_driver external IO writing.

[gavanderhoorn]

From the DX200 manual:

So that would be [27010; 29567] for Network Inputs and [10010; 15127] for General Outputs.

Difficulty would be to "know" which controller is on the other end. As for FS100/DX100 these numbers are different.

[ted-miller]

DX100: 10010 - 12567 25010 - 27567

FS100: 10010 - 11287 25010 - 26287

DX200: listed in your image above YRC1000 and micro: Same as DX200.

[ted-miller]

Also, keep in mind that the last digit will never exceed 7.

10010 - 10017 = Valid 10018 - 10019 = NOT VALID 10020 - 10027 = Valid

[gavanderhoorn]

@ted-miller: thanks.

@marip8 @steviedale: if/when we introduce the stand-alone IO relay we can provide it with the information on the connected controller, so as to filter for the correct ranges.

[gavanderhoorn]

@ted-miller: the DX200 Reference Manual For New Language Environment MotoPlus (166580-1CD) suggests the interface panel IOs are also writable:

I haven't checked for the other controllers, but are they similar?

[steviedale]

@gavanderhoorn @ted-miller these address values are all in decimal right (not hex)?

[steviedale]

@gavanderhoorn as far as identifying which controller is being used, one possible solution would be to create a separate io_relay.launch file for each type, then have each provide a ROS param that contains the name of the controller. What do you think?

[gavanderhoorn]

Please check the discussion in #367. It may be more efficient to extend MotoROS to return appropriate error codes, and then add generic error handling code to the io_relay.

[ted-miller]

Please check the discussion in #367. It may be more efficient to extend MotoROS to return appropriate error codes, and then add generic error handling code to the io_relay.

See PR #369

[gavanderhoorn]

This is addressed (in a slightly different way) by #371.