Recently we developed a novel algorithm for simulating hard objects of arbitrary shape, decorated on their surface with spherical sites, that interact through an attractive step potential. Using this new algorithm, we simulated two recently studied primitive models of network forming liquids — a primitive model for water and for silica. We evaluated the isodiffusivity lines in the temperature-density plane to provide an indication of the shape of the glass transition line. Except for large densities, arrest is driven by the onset of the tetrahedral bonding pattern (i.e. the bonding energy controls arrest) and the resulting dynamics is strong in Angell's classification scheme. Notably patchy colloidal particles of new generation may designated to closely resemble the primitive models of Silica and Water, thus it is intriguing to think that this arrest mechanism driven by bonding energy, at low volume fractions, is the same that controls the gel formation, entitling us to call glasses all arrested states, including the gel states.