(up to isomorphism)
(if my latex were better i'd try to put it all on one line, this is a right-split short exact sequence).
if so, is there a "smaller" diagram which conveys the same information?
yes i did mean the other composition ( can i claim it was because i was looking at herstein last night? no? ok, i admit it, i'm just stupid).
so...how would i indicate there is no left-split with a picture, to rule out the trivial semi-direct (that is direct) product?
correct. what got me thinking about this, is thinking about identifying D4 solely in terms of homomorphisms to and from it (although i admit including the extra information of the short exact sequence is cheating a little) and wondering how well this can be done, for an arbitrary group.