Closed jslee02 closed 5 years ago
I think URDF wasn't designed to support multi-tree robot descriptions, so the original author of this (which wasn't me; all I did was refactor this code) decided that we should consider instances of multi-tree URDFs to be an error.
There are a few things about our URDF parser that aren't really standard, though, so I'm completely in favor of supporting a superset of the URDF standard.
I think URDF wasn't designed to support multi-tree robot descriptions
That is true. So I think we should either return nullptr
with an error message for that case (which is not the current implementation) or allow multi-tree robot with a warning message saying it's not the standard.
I'm inclined to the second option.
The second option sounds good to me too :+1:
Resolved by #1270
URDF parser errors when the root link is "world" (treated as a keyword) and the root link has multiple children. However, DART supports multitree in a skeleton. Moreover, this logic continues with this error and yet surprisingly it successfuly creates multitree anyway.
We should make it clear that it's not an error and update the logic accordingly.