In the above paper, a composite energy function learning control approach was proposed to asymptotically eliminate the mismatch between the desired and actual periodic trajectory of a system. Upon review of this result, there appear to be several philosophical and technical issues that invalidate the result including the use of a resetting condition and the lack of boundedness of the learning estimate. The intent of this comment is to highlight these technical errors, especially since the boundedness of the learning estimate has historically been a problematic issue.