inla.stack.e currently joins exposures E for multiple likelihoods into a matrix, just as for the observables, but it should be a vector (each row of the observable is attached to only one likelihood model, so there is no ambiguity). (also see #33) The inla() behaviour when given an E matrix is unknown and undefined.
Ntrials for the binomial likelihood needs to be handled in the same way as E, and be a parameter to like().
inla.stack.e currently joins exposures E for multiple likelihoods into a matrix, just as for the observables, but it should be a vector (each row of the observable is attached to only one likelihood model, so there is no ambiguity). (also see #33) The inla() behaviour when given an E matrix is unknown and undefined.
Ntrials for the binomial likelihood needs to be handled in the same way as E, and be a parameter to like().