Writing a custom model for gnm package, but need a function not in derivative table. Any workarounds?

I am trying to write a custom model for gnm, to fit Voigt profiles. I have defined

library(gnm) ; library(RcppFaddeeva)
VOIGT <- function(x) {
    list(predictors=list(Center=1, Sigma=1, Gamma=1, Alfa=1),
         variables=list(substitute(x)),
         term=function(predLabels, varLabels) {
        paste(predLabels[4], "* RcppFaddeeva::Voigt(", varLabels[1], ", ", predLabels[1], ", ", predLabels[2], ", ", predLabels[3], ")" )
    },
    start=function(theta) {
     theta[2] <- 1 ; theta[3] <- 1; theta[4] <- 1
        } )
}

class(VOIGT) <- "nonlin"

but trying to use it gives

Error in deriv.default(X[[i]], ...) : 
  Function 'RcppFaddeeva::Voigt' is not in the derivatives table

Any workarounds? Or does this need extensions to gnm, or some completely other approach?