The release notes of NetLogo 7.0.4 state “BehaviorSpace was not giving early or informative errors, when the world dimension parameters were clearly invalid”.
Firstly, thank you for addressing this issue. This is an important improvement.
That being said, I think there is still room for improvement.
It would be helpful if the message includes the offending value, in particular for world-width and world-height.
The error message on wrong values for min-pxcor, max-pxcor, min-pycor, and max-pycor seems a bit odd. It says “One of this experiment’s parameter combinations has conflicting values …”, while the problem is just min-xpcor or min-pycor being positive or max-pxcor or max-pycor being negative, independently of the corresponding max/min value. (As far as I could find out, there is no check on keeping one direction centered if it was centered, which would be problematic anyway.)
For checking values of world-width and world-height, it is debatable which checks are useful, apart from a test on being strictly positive.
The model itself could change the world boundaries between runs (not advisable, but not prohibited either). Then if the first combination of world-width and world-height is accepted, there is no way to tell if width/height values for later runs are valid or not.
Flagging an even value as invalid could be too strict, but accepting an odd value could still lead to invalid resizing.