[R] Checking for a proper "stop" statement...

[Duncan Murdoch]

On 21/02/2018 4:28 PM, Jonathan Greenberg wrote:
> Folks:
> 
> Consider the following two use cases:
> 
> goodfunction <- function()
> {
> stop("Something went wrong..."
> }
> 
> # vs.
> 
> badfunction <- function()
> {
> notgood()
> }
> 
> Is there a way for me to test if the functions make use of a stop()
> statement WITHOUT modifying the stop() output (assume I can't mod the
> function containing the stop() statement itself)?  For "goodfunction" the
> answer is TRUE, for "badfunction" the answer is FALSE.  Both return an
> error, but only one does it "safely".
> 
> I thought the answer might lie in a tryCatch statement but I'm having a
> hard time figuring out how to do this test.
> 

I think tryCatch() is what you want.  To see the difference between 
those two errors, run

tryCatch(goodfunction(), error = function(e) browser())

and similarly with badfunction().  When you land in the browser, you'll see

Browse[1]> str(e)
List of 2
  $ message: chr "Something went wrong..."
  $ call   : language goodfunction()
  - attr(*, "class")= chr [1:3] "simpleError" "error" "condition"

and

Browse[1]> str(e)
List of 2
  $ message: chr "could not find function \"notgood\""
  $ call   : language notgood()
  - attr(*, "class")= chr [1:3] "simpleError" "error" "condition"


so you could write a test based on the message, or based on the call to 
see where the stop() happened.

Duncan Murdoch