r：函数参数先匹配变量，然后再省略号

Question

here is my function:

print.log <- function(print.flag = T,...){
  if(print.flag)
    print(paste(as.character(Sys.time()),' - ',...))
}

function call:

print.log('listing x files')

error message: Error in if (print.flag) print(paste(as.character(Sys.time()), " - ", : argument is not interpretable as logical

here is the value print.flag holds:

Browse[2]> print.flag
[1] "listing x files"

I understand that the string I wanted to push to the ellipsis was matched after all to the print.flag variable.

my question: why is that the behavior, and how can I fixed it to achieve the desired result? (meaning, all function arguments will go to the ellipsis, unless I specifically declare the print.flag variable )

Answer 1

The solution here is actually pretty easy: switch the order of your arguments.

print.log <- function(..., print.flag = T){
  if(print.flag)
    print(paste(as.character(Sys.time()),' - ',...))
}

The reason this works is that unnamed parameters in a function call are matched in the order they are named in the function definition. This happens until you hit ... , which greedily eats up all unnamed (or named and not mentioned, for that matter) arguments.

print.log("a", "b", "c")
#> [1] "2017-07-30 05:21:08  -  a b c"

print.log("a", print.flag = F)

print.log("a", print.flag = T, "b")
#> [1] "2017-07-30 05:21:08  -  a b"

print.log("a", one = "b", two = "c")
#> [1] "2017-07-30 05:21:08  -  a b c"

NOTE: perhaps to answer a few questions you did not ask.

Be careful using . as a naming convention (despite many base R functions that do not follow this advice), as the . is used by S3 for class behavior. More reading here .

Also, you might look into the logging package, which is an R port of the popular python logging package. It allows environmental control of logging with "logging level" and defining a default formatting (in your case, timestamp in front, etc.).