I’ve recently started coding in Python more than I’ve used to, and it’s interesting for me to have a closer look to its features and compare them with similar Julia’s ones.

The do-block in the Julia programming language and the with statement in Python can appear to be similar because of some common uses (most notably, opening a stream and automatically closing it at the end), but are fundamentally different. In this post I will review them.

## with statement in Python

The with statement has been introduced in Python with PEP 343 to simplify the control flow of try/finally statements.

A good explanation of this feature is given at http://effbot.org/zone/python-with-statement.htm. In general, you can use try/finally to handle unmanaged resources like streams:

def controlled_execution(callback):
set things up
try:
callback(thing)
finally:
tear things down

def my_function(thing):
do something with thing

controlled_execution(my_function)


For an object of a class, it is possible to reorganize the above control flow by using the with statement:

class controlled_execution:
def __enter__(self):
set things up
return thing
def __exit__(self, type, value, traceback):
tear things down

with controlled_execution() as thing:
do something with thing


The __enter__ method prepares the object for later use and returns thing, which can be then consumed in the body of with. The __exit__ method will be called in any case at the end of the body of with and can be used to close the previously opened resources and handle any exception raised within the with body, if necessary.

### Opening a file with with

A common application of the with statement is the open function:

with open(filename, mode='w') as f:
do something with f


Just to practically see how the __enter__ and __exit__ methods can be defined, we implement a very simple class, MyFile, to replicate the behaviour of the builtin open function when opening a file:

class MyFile:
def __init__(self, file):
self.file = file
def __enter__(self):
return self.file
def __exit__(self, type, value, traceback):
return self.file.close()

def MyOpen(name, mode='r'):
return MyFile(open(name, mode))

with MyOpen('file.txt') as f:


## do-block in Julia

The do-block in Julia addresses a different problem: write in a different way the passing of a function as argument to another function.

In a simple way, any function func accepting a function as first argument:

function func(f::Function, x, y, z...)
# do something with arguments x, y, z..., and call f at some point
end


can be called as

func(x, y, z...) do args
# put here the body of function f(args) that will
# be passed as first argument to func(f, x, y, z...)
end


For example, the filter enables filtering a collection by removing those elements for which the provided function returns false. We can select the odd numbers between 1 and 100 divisible by 3 with

julia> filter(x -> isinteger(x/3) && isodd(x), 1:100)
17-element Array{Int64,1}:
3
9
15
21
27
33
39
45
51
57
63
69
75
81
87
93
99


Here we passed an anonymous function as first argument to filter in order to select the desired elements of the collection passed as second argument. We can rewrite this by using the do-block in the following way:

julia> filter(1:100) do x
isinteger(x/3) && isodd(x)
end
17-element Array{Int64,1}:
3
9
15
21
27
33
39
45
51
57
63
69
75
81
87
93
99


The code in the body of the do-block is used to create an anonymous function, whose arguments are specified after do, that is passed as first argument to the function before the do.

### Opening a file with do

A common use of the do-block is to open a file and automatically close it when you’re done, just like the with statement in Python. For example:

open("outfile", "w") do io
write(io, data)
end


In Julia’s Base code, this is achieved by extending the open method in the following way:

function open(f::Function, args...; kwargs...)
io = open(args...; kwargs...)
try
f(io)
finally
close(io)
end
end


This method, which accepts a function as first argument, is what makes it possible to use the do-block in the example above: after effectively opening the file, the body of the do-block is executed in the try statement, and in the end the finally clause takes care of safely closing the file, even in case of errors while executing the function f.

## Summary

The with statement in Python is a syntactic sugar for try/finally. This construct is bound to classes, this means that the __enter__/__exit__ methods are bound to a specific class and they can be defined only once as they have a fixed signature, so you can’t have multiple __enter__/__exit__ handlers within the same class, or independently from a class. In order for the with statement to work, both __enter__ and __exit__ must be defined in any case, however if you don’t really need the try/finally construct you can still take advantage of the handy with syntax, for example by making the __exit__ method dummy.

The do-block in Julia is not bound to any type, the only condition to use it is to have a function taking a function as first argument. The function may contain a control flow construct, like try/finally, but this is not mandatory. You can also have different methods for a function to be used with the do-block, depending on the types and/or the number of the other arguments.