An acid anhydride is condensed with an aromatic aldehyde in the Perkin reaction. The latter is non-enolisable, and thus the former forms the enolate which attacks as a nucleophile. If the attack is made at the aromatic aldehyde, an aldol-like product is formed which can be hydrolysed to the acid.
However if the acid anhydride enolate attacks another acid enolate, in a nucleophilic addition-elimination step, this results in the leaving of a carboxylate ion and the generation of an active methylene group (similar to Claisen condensation). There may be proton transfer between the two later. Surely this is more favourable than the first mechanism? Why doesn't this occur though?
Similarly, in the Stobbe condensation, instead of attack by the enolate at the ketone or aldehyde, cannot the succinic ester be attacked again?