case of bad error trapping for redefined variable in model

[paciorek]

We should be doing a better job error trapping here.

 code <- nimbleCode({
    for(i in 1:5) {
        y[i] ~ dnorm(b0+x[i]*b1, sd = sigma[i])
                                        # log(sigma[i]) <- a0 + a1*x[i]
        sigma[i] <- exp(a0+a1*x[i])
    }
    sigma~dhalfflat()
    b1~dnorm(0,1)
    b0~dnorm(0,1)
    a1~dnorm(0,1)
    a0~dnorm(0,1)
})

model=nimbleModel(code,data=list(y=rep(0,5)))
Defining model
Error in 1:dims[1] : argument of length 0