Variable Scope
In this lesson we will learn what is the scope of a variable and how scopes can be used to rule when a variable should be accessible in our program.
The scope of a variable is the region of a program where the variable is known and accessible. A variable may live in two kind of scopes: the global scope or a local scope.
Scope
A variable in the global scope is accessible everywhere in the program and can be modified by any part of the code. When we define a variable in the REPL or outside of a function, for example, we create a global variable.
A variable in a local scope is only accessible in that scope and in other scopes eventually defined inside it.
In Julia there are several constructs which introduces a scope:
Construct | Scope type | Scope blocks it may be nested in |
---|---|---|
module , baremodule |
global | global |
interactive prompt (REPL) | global | global |
(mutable) struct , macro |
local | global |
for , while , try-catch-finally , let |
local | global or local |
functions (either syntax, anonymous & do-blocks) | local | global or local |
comprehensions, broadcast-fusing | local | global or local |
As you can see, some constructs introduce a global scope (for example each module has its separate global scope) and others introduce a local scope (for example functions
, for
loops and let
blocks).
Let’s now see in more details how to work with scopes and in which scope it is better to define a variable, depending on its usage.
Local Scope
Let’s start with a construct we are already familiar with: a function. A function declaration introduce a scope, which means that each variable declared inside a function lives inside the function: it can be accessed freely inside the function but cannot be accessed outside the function, which is good! Thanks to this property we can use the names most suitable for our variables (x, y, z, etc.) without the risk of clashing with the declaration of other functions.
If a value computed inside a function is needed outside the function it is a good idea to return that value instead of modifying a global constant external to the function. As a rule of thumb, a function should rely only on its input parameters and return only the variable which may be useful for further computations.
Let’s see an example of a variable which exist inside a function (local scope) but doesn’t exist in the global scope:
function example1()
z = 42
return
end
>>> z
ERROR: UndefVarError: z not defined
Although it is not advisable, it is possible to specify that a variable should be accessed in the global scope through:
function example2()
global z = 42
return
end
>>> example2()
>>> z
42
A better approach is instead to return z and let the user perform the allocation of z:
function example3()
z = 42
return z
end
>>> z = example3()
>>> z
42
In the case where it is necessary to distinguish between a variable which exists both in the local and global scope, it is possible to indicate the one that needs to be used through the global
or local
annotation before the desired variable (for more details see this page of the Julia documentation).
let
construct
The let
construct can be used to introduce a new local scope. It is useful, for example, when you want to perform some computations but you don’t want the intermediate results/variables to pollute your current scope.
The let block will be able to access all the local (or global) variables available in its parent scope and will have its own set of local variables. It is also possible to specify some initial values to mimic the execution of a function once.
a = let
i=3
i+=5
i # the value returned from the computation
end
>>>a
8
b = let i=5
i+=42
i
end
>>>b
47
c = let i=10
i+=42
i
end
>>>c
52
>>>i
UndefVarError: i not defined
As you can see, let
blocks are pretty convenient when it comes to splitting computations over several lines, but there are other possible uses as shown here.
let
blocks are somewhat similar to begin
blocks, although begin
blocks don’t introduce a local scope:
d = begin
i=41
i+=1
i
end
>>>d
42
>>>i
42
Global scope
Whenever we define a variable in the REPL or in general outside a construct which introduces a local scope, we place a variable in the global scope. The global scope is accessible everywhere in the program and a variable in the global scope can be modified by any part of the code. As such, it is generally advisable to avoid using global variables as much as possible, in fact since global variables can change their type and value at any time, they cannot be properly optimised by the compiler.
Constants
One way to mitigate this performance issue is to define global variables as constants through the const
annotation. When we think of a constant we generally imagine a variable which is immutable, for example the speed of light: the speed of light is a number and so it could be expressed by a Float64
, there is no need to change its type, once it has been defined, to a String
, in this case.
A constant in Julia is a variable which cannot change its type once it is defined (Julia will throw an error if there is an attempt to modify the type of a constant) and Julia will show a warning message if we try to modify the value of a constant. Since a constant is “type immutable” it can be properly optimised by the compiler and there are fewer performance issues.
>>> const C = 299792458 # m / s, this is an Int
299792458
>>> C = 300000000 # change the value of C
WARNING: redefining constant C
>>> C = 2.998 * 1e8 # change the type of C, not permitted
invalid redefinition of constant C
Modules
If you don’t already know what a module is you don’t have to worry, we will talk about modules in the next lessons and you can come back to this part later!
As we have anticipated, modules introduce separate global scope, which means that a variable which is known inside a module will not be accessible outside of the module unless it is exported or it is accessed through the ModuleName.varName
notation.
module ScopeTestModule
export a1
a1 = 25
b1 = 42
end # end of module
using .ScopeTestModule
>>>a1
25
>>>b1
UndefVarError: b1 not defined
>>>ScopeTestModule.b1
42
>>>ScopeTestModule.b1=26
cannot assign variables in other modules
At line 9-10 we can see that the a1
variable, which is exported by the module, can be accessed directly without specifying the scope of the variable, while on line 12-13 and 15-16 we can see that b1
can only be accessed by specifying where the variable lives (i.e. inside ScopeTestModule
). At line 18-19 we can see that it is not possible to directly modify a variable which is defined inside another module.
Conclusions
When you are writing some quick computations in the REPL or you are writing a simple script, you don’t need to care about what goes in the global scope and what not, but if you are writing a piece of code that has to be reusable (like a library for example) and needs to be fast, there are some guidelines which should be followed:
-
Avoid using global variables as much as possible, if global variables are needed define them as
const
-
Pass all the required parameters to a function instead of defining them as global variables, a function should be able to ideally operate only on its input.
-
Use functions and let blocks to introduce local scopes where you can define as many variables as you desire without incurring in the risk of overlapping variable names with those used in other parts of your code.
If you liked this lesson and you would like to receive further updates on what is being published on this website, I encourage you to subscribe to the newsletter! If you have any question or suggestion, please post them in the discussion below!
Thank you for reading this lesson and see you soon on TechyTok!
Leave a comment