Re: Construct a function with argments specified by an R expression

Another way to do this is to make a prototype and then modify it.  For 
example,

   foo <- function(x, y, others) {
     # the body can be anything, we won't touch it
   }

   args <- formals(foo)

   # Make extra copies of the 3rd argument
   args[3:(2 + length(argnms))] <- args[3]

   # Set the names of the new ones (or use Iris's code to set all the 
names, your choice)

   names(args)[3:(2 + length(argnms))] <- argnms

   formals(foo) <- args

Duncan Murdoch

On 2025-10-24 11:17 p.m., Rolf Turner wrote:
> 
> On Fri, 24 Oct 2025 22:31:53 -0400
> Iris Simmons <[email protected]> wrote:
> 
>> I would start by making a list of the desired length, where each
>> element is the missing argument:
>>
>> f <- rep(list(quote(expr = )), 2 + length(argnms))
>>
>> then add names to the list:
>>
>> names(f) <- c("x", "y", argnms)
>>
>> and then use formals<-
>>
>> formals(foo) <- f
> 
> Wow! That was fast!  And it seems to do exactly what I want.
> 
> What I was missing, I guess, was the quote(expr = ) structure.  I have
> never really understood quote(), nor expr() either.  I'll just treat it
> as a black box for now.
> 
>> Although, I would ask if you're certain that this is actually what
>> you want to do. It's usually a bad idea to have variable names that
>> you don't know. I'd to think it'd be better to have the default
>> argument:
>>
>> a = list()
>>
>> and supply a list of `a` values instead of several `a` arguments.
> 
> Yes I understand that that would be the "sensible" approach but it
> doesn't work in the context with which I am confronted.
> 
> I have more work to do before I get the approach that I've envisaged up
> and running.
> 
> I may inflict more questions upon you later.  (No good deed goes
> unpunished! 🙂️)
> 
> 
> cheers,
> 
> Rolf
>