The Call Stack

Estimated time: 20 minutes

Overview

Questions

• What is the call stack, and how does R know what order to do things in?
• How does scope work in R?

Objectives

• Explain how stack frames are created and destroyed as functions are called.
• Correctly identify the scope of a function’s local variables.
• Explain variable scope in terms of the call stack.

The Call Stack

Let’s take a closer look at what happens when we call fahrenheit_to_kelvin(32). To make things clearer, we’ll start by putting the initial value 32 in a variable and store the final result in one as well:

R

original <- 32
final <- fahrenheit_to_kelvin(original)

The diagram below shows what memory looks like after the first line has been executed:

When we call fahrenheit_to_kelvin, R doesn’t create the variable temp_F right away. Instead, it creates something called a stack frame to keep track of the variables defined by fahrenheit_to_kelvin. Initially, this stack frame only holds the value of temp_F:

When we call fahrenheit_to_celsius inside fahrenheit_to_kelvin, R creates another stack frame to hold fahrenheit_to_celsius’s variables:

It does this because there are now two variables in play called temp_F: the argument to fahrenheit_to_celsius, and the argument to fahrenheit_to_kelvin. Having two variables with the same name in the same part of the program would be ambiguous, so R (and every other modern programming language) creates a new stack frame for each function call to keep that function’s variables separate from those defined by other functions.

When the call to fahrenheit_to_celsius returns a value, R throws away fahrenheit_to_celsius’s stack frame and creates a new variable in the stack frame for fahrenheit_to_kelvin to hold the temperature in Celsius:

It then calls celsius_to_kelvin, which means it creates a stack frame to hold that function’s variables:

Once again, R throws away that stack frame when celsius_to_kelvin is done and creates the variable temp_K in the stack frame for fahrenheit_to_kelvin:

Finally, when fahrenheit_to_kelvin is done, R throws away its stack frame and puts its result in a new variable called final that lives in the stack frame we started with:

This final stack frame is always there; it holds the variables we defined outside the functions in our code. What it doesn’t hold is the variables that were in the various stack frames. If we try to get the value of temp_F after our functions have finished running, R tells us that there’s no such thing:

R

temp_F

ERROR

Error in eval(expr, envir, enclos): object 'temp_F' not found

Callout

Where to Learn More

The explanation of the stack frame above was very general and the basic concept will help you understand most languages you try to program with. However, R has some unique aspects that can be exploited when performing more complicated operations. We will not be writing anything that requires knowledge of these more advanced concepts. In the future when you are comfortable writing functions in R, you can learn more by reading the R Language Manual or this chapter from Advanced R Programming by Hadley Wickham. For context, R uses the terminology “environments” instead of frames.

Why go to all this trouble? Well, here’s a function called span that calculates the difference between the minimum and maximum values in an array:

R

span <- function(a) {
  diff <- max(a) - min(a)
  return(diff)
}

dat <- read.csv(file = "data/inflammation-01.csv", header = FALSE)
# span of inflammation data
span(dat)

OUTPUT

[1] 20

Notice span assigns a value to variable called diff. We might very well use a variable with the same name (diff) to hold the inflammation data:

R

diff <- read.csv(file = "data/inflammation-01.csv", header = FALSE)
# span of inflammation data
span(diff)

OUTPUT

[1] 20

We don’t expect the variable diff to have the value 20 after this function call, so the name diff cannot refer to the same variable defined inside span as it does in the main body of our program (which R refers to as the global environment). And yes, we could probably choose a different name than diff for our variable in this case, but we don’t want to have to read every line of code of the R functions we call to see what variable names they use, just in case they change the values of our variables.

The big idea here is encapsulation, and it’s the key to writing correct, comprehensible programs. A function’s job is to turn several operations into one so that we can think about a single function call instead of a dozen or a hundred statements each time we want to do something. That only works if functions don’t interfere with each other; if they do, we have to pay attention to the details once again, which quickly overloads our short-term memory.

Challenge

Following the Call Stack

We previously wrote functions called highlight and edges. Draw a diagram showing how the call stack changes when we run the following:

R

inner_vec <- "carbon"
outer_vec <- "+"
result <- edges(highlight(inner_vec, outer_vec))

Show me the solution

R evaluates this expression from the inside out. Reading from left to right, the call stack passes through five distinct states:

1. Global frame only — inner_vec and outer_vec are bound in the global environment. The call stack contains a single frame:

   [global]  inner_vec = "carbon"
             outer_vec = "+"

2. Entering highlight — R needs the first argument of edges, so it must call highlight(inner_vec, outer_vec) first. A new stack frame is pushed for highlight. Inside that frame, the formal arguments (let’s call them content and wrapper) bind to the caller’s inner_vec and outer_vec values:

   [highlight]  content = "carbon"
                wrapper = "+"
   [global]     inner_vec, outer_vec

3. highlight returns — the function returns its result (e.g., "+carbon+"). R pops the highlight frame; its locals (content, wrapper) are gone:

   [global]  inner_vec, outer_vec
             [returned value: "+carbon+"]

4. Entering edges — R now calls edges("+carbon+"). A new frame is pushed for edges, with its formal argument bound to the returned string:

   [edges]   v = "+carbon+"
   [global]  inner_vec, outer_vec

5. edges returns; assignment to result — edges produces its return value (e.g., c("+", "+")). The edges frame is popped, and the return value is bound to result in the global environment:

   [global]  inner_vec = "carbon"
             outer_vec = "+"
             result    = c("+", "+")

The two important takeaways are:

• Frames are pushed in evaluation order, not lexical order. Even though edges(...) appears outermost in the source code, highlight(...) is pushed first because R must evaluate the argument before the call.
• Local variables die with their frame. If you try to print content, wrapper, or v at the global prompt, R reports “object not found” — those names existed only inside their respective frames.

A handy way to confirm this in practice is to insert traceback() inside one of the helper functions, or to set options(error = recover) before triggering an error inside highlight — R will list the active frames and let you inspect their locals.

Key Points

• R keeps track of active function calls using a call stack comprised of stack frames.
• Only global variables and variables in the current stack frame can be accessed directly.