Hello, Has the notion of Cauchy-completion of a bicategory been considered? It looks to me that instead of looking at arbitrary idempotent 1-cells f together with an isomorphism \alpha : ff ---> f one has to look at those pairs <f,\alpha> where \alpha is associative. Without this requirement I seem to run into coherence problems. Any pointers to the literature would be welcome. Thank you! -- J"urgen Koslowski Institut f"ur theoretische Informatik TU Braunschweig Braunschweig, Germany