[R] protecting "..."

[Duncan Murdoch]

Johannes Hüsing wrote:
> Duncan Murdoch <murdoch at stats.uwo.ca> [Sat, Nov 17, 2007 at 11:56:52PM CET]:
>   
>> On 17/11/2007 5:04 PM, Johannes Hüsing wrote:
>>     
>>> I just realized that the "..." argument in a function cannot be used 
>>> without
>>> taking precautions sometimes. The following behaviour is what I 
>>> stumbled upon:
>>>       
>> The problem is the way replicate is implemented.  It takes the 
>> expression, and wraps it in a function(...) before calling sapply.  The 
>> ... in your expression matches the ... in the wrapper function, not the 
>> ... from the parent.
>>     
>
> Many thanks! Pose a question, go to sleep, wake up, read the answer.
> I feel like being in old Cologne 
> (http://en.wikipedia.org/wiki/Heinzelm%C3%A4nnchen).
>
>
>   
>> I think you'll need to make calls to sapply directly to do what you 
>> want.
>>
>>     
>
> Do you mean to replace "replicate" by the contents of the function?
> I am feeling more comfortable with my solution.
What I meant was something like this:


myrepl2 <- function(length, fun, ...) {
 iterfun <- function(n) fun(...)
 sapply(integer(length), iterfun) 
}

but your solution works too.

Duncan Murdoch