Once requirements are fully understood, it is important to have them documented and verified. This stage is critical, since no one wants to waste hundreds of employee hours developing a software package that does not function as expected. Requirements specification can be done in terms of the logical components involved, their respective inputs, outputs and desired relationships, and constraint to be imposed. States and possible state transitions can also be incorporated into a requirements specification; in this case the requirements specification may be executable. Indeed some requirements specification languages are executable (e.g., Petri net; see Section 3.6).