R - How to handle the dot-dot-dot (ellipis/"...") with multiple subsequent functions - i.e. passing only some of the variables

Question

I'm working on a non-linear optimization, with the constrOptim.nl from the alabama package. However, my problem is more related to passing arguments (and the dot-dot-dot (ellipis/"...") and maybe do.call)- so I give first a general example and later refer to the constrOptim.nl function.

Suppose, I have the following functions - from which I only can edit the second and third but not the first.

first<-function (abc, second, third, ...){
second(abc,...)
third(abc,...)
}

second<- function(abc, ttt='nothing special'){
print(abc)
print(ttt)
}

third<- function(abc, zzz="default"){
print(abc)
print(zzz)
}

The output I want is the same I would get when I just run

second("test", ttt='something special')
third("test", zzz="non-default")

This is

"test"
"something special"
"test"
"non-default"

However, the code below doesn't work to do this.

first("test",second=second, third=third, ttt='something special',zzz="non-default")

How can I change the call or the second and third function to make it work?

http://www.r-bloggers.com/r-three-dots-ellipsis/ here I found some advice that do.call could help me but at the moment I'm not capable of understanding how it should work.

I cannot change the first function since this is the constrOptim.nl in my particular problem - and it is designed to be capable of passing more arguments to different functions. However, I can change the second and third function - as they are the restrictions and the function that I want to minimize. Obviously I can also change the call of the function.

So to be more particular, here is my specific problem: I perform a maximum likelihood estimation with non-linear restrictions:

minimize <- function(Param,VARresiduals){ 
   #Blahblah
   for (index in 1:nrow(VARreisduals)){
   #Likelihood Blahbla
   }
  return(LogL)
}
heq<-function(Param,W){   
  B<-Param[1:16]
  restriction[1]<-Lrestriction%*%(diag(4)%x%(solve(W))%*%as.vector(B))
  restriction[2:6]<-#BlablaMoreRestrictions
  return(restriction)
}

Now I call the constrOptim.nl...

constrOptim.nl(par=rnorm(20), fn=minimize,hin=NULL heq=heq,VARresiduals,W)

...but get the same error, as I receive when I call the first function above - something like: "Error in second(abc, ...) : unused argument (zzz = "non-default")".

How can I change minimize and heq or the call? :) Thanks in Advance

Update after the post got marked as a duplicate: The answer to the related post changes the first function in my example - as it implements a do.call there, that calls the other functions. However, I cannot change the first function in my example as I want to keep the constrOptim.nl working a variety of different functions. Is there another way?

Answer 1

The solution I came up with is not very elegant but it works.

second_2<- function(abc, extras){
  a<-extras[[1]]
  print(abc)
  print(a)
}

third_2<- function(abc, extras){
  a<-extras[[2]]
  print(abc)
  print(a)
}

extras<-list()
extras[[1]]<-'something special'
extras[[2]]<-"non-default"
extras


first("test",second=second_2, third=third_2, extras)

surprisingly also the following code works, but with a slightly different outcome

 first("test",second=second, third=third, extras) 

after all, setting default values is now a little clumsy but not infeasible.