Should carbohydrates be constrained in an aqueous simulation box?

GROMACS version: 2023
GROMACS modification: Yes/No
Any opinions or papers indicating a need to constrain the angles and bond lengths of carbohydrate molecules in a water-carbohydate-protein simulation?