fix prepExtendedComment called from package::fun

[pascal-sauer]

No description provided.

[pascal-sauer]

fixes #226

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

A bit unsafe to look for the function in some environment when functionCall[[2]] explicitly states it. But this is an internal function, so OK.
But why not

functionCall <- tail(as.character(cl[[1]]), 1)

?

[pascal-sauer]

A bit unsafe to look for the function in some environment when functionCall[[2]] explicitly states it

I assume you refer to the else branch? This is the behavior we had previously which I did not see a reason to change.

functionCall <- tail(as.character(cl[[1]]), 1)

I assume you suggest to skip the if else and use this instead to then call get(functionCall, mode = "function", sys.frame(-n - 1))? This won't work in a scenario where madrat::readSource is called, unless madrat is also attached (in which case readSource likely won't be called with the madrat:: prefix), because get would only look on the search path.

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

The pos argument can specify the environment in which to look for the object in any of several ways: as a positive integer (the position in the search list); as the character string name of an element in the search list; or as an environment (including using sys.frame to access the currently active function calls). The default of -1 indicates the current environment of the call to get. The envir argument is an alternative way to specify an environment.

—https://rdrr.io/r/base/get.html

The fully qualified call is

f <- get(functionCall[[3]], 
         envir = loadNamespace(functionCall[[2]]),
         mode = "function",
         pos = sys.frame(-n - 1))

Now, which argument is used, envir or pos (as they point to different things)?

Spoiler:

It is envir.

Since prepExtendedComment() is not exported, people calling it without attaching madrat are wholly responsible for their own pain ;).

But a safe formulation would be

f <- get(x = tail(functionCall, 1),
         envir = ifelse(1 == length(functionCall),
                        sys.frame(-n - 1),                # function() call
                        getNamespace(functionCall[[2]])), # pkg::function() call
         mode = "function")

[0UmfHxcvx5J7JoaOhFSs5mncnisTJJ6q]

(I assumed – wrongly – that pos would take precedence over envir and that get(…, sys.frame(-n - 1), inherits = TRUE) would just climb up the search path until some matching function is found.)

[pascal-sauer]

I found a way to get rid of this call stack inspection altogether