#decrease len if you have little memory.
predplot <- function(object, data, class, len = 100, ...)
{
    plot(data,col=unclass(class),pch=unclass(class))
    xp <- seq(min(data[,1]),max(data[,1]),length=len)
    yp <- seq(min(data[,2]),max(data[,2]),length=len)
    grid <- expand.grid(xp,yp)
    colnames(grid) <- colnames(data)
    Z <- predict(object, grid, ...); 
    zc <- as.numeric(Z$class)
#browser()
    for (i in unique(zc))
	{zp <- Z$post[,i] - apply(as.matrix(Z$post[,-i]),1,max)
        contour(xp, yp, matrix(zp, len),
            add = T, levels = 0, drawlabels = F)}
    invisible()
}