Any comments?
Any fool can write code that a computer can understand. Good programmers write code that humans can understand.
— Martin Fowler, 20191
Great code must balance various tensions. It should be comprehensible to a human reader as well as being self-documenting. You and I are not coding geniuses (yet). We should strive to document the Haskell code we write and document it well.
Single-line comments
A single-line Haskell comment starts with the double dash symbol --
. The “Hello, world” program from the section Tales from the script can be documented like so:
1
2
-- The traditional greeting.
main = print "Hello, world"
The double dash --
can be used for multiline comments. All you need to do is prefix each comment line with --
. Haskell offers another way to insert multiline comments.
Multiline comments
Some programming languages offer a way to comment multiple lines.2 Haskell is no exception. The beginning of a multiline Haskell comment is marked by the symbol {-
. The end of the multiline comment is designated by -}
. Here is the “Hello, world” program accompanied by a multiline comment.
1
2
3
4
5
{-
This is the first program you would write when you learn a programming
language. This traditional program is called the "Hello, world" program.
-}
main = print "Hello, world"
When in Rome…
The GHC project has its own tool to automatically extract documentation from Haskell source files. The utility expects you to document your Haskell code using various annotations. The very first documentation line of a function is annotated as -- |
, i.e. the double dashes followed by a whitespace and the pipe symbol. Each subsequent line of documentation is prefixed by --
. Here is the “Hello, world” program annotated according to the Haskell way of documentation.
1
2
3
4
5
-- | This is the first program you would write when you learn a
-- programming language. This traditional program is called the
-- "Hello, world" program. This documentation style is used by the GHC
-- project to document Haskell source code.
main = print "Hello, world"
Why document?
A computer will ignore your comments. Why bother inserting comments in your Haskell source files? The usual reason is documentation. A computer executes your Haskell code, but it is a human (most probably yourself) who would be reading and trying to understand the code at some future time. You write code not just for a computer, but for other humans as well.
Imagine you are trying to solve a programming problem. You struggle for a few hours, maybe a few days or weeks. After a while, you find a clever solution and implement it in Haskell code. The code works and outputs the correct result. You move on to another problem or some other challenge. Some time passed, probably a few weeks or months. You come back to the code for the clever solution and have forgotten how or why the code works. Reading the code gives you some hints, but you still struggle to remember the name of the technique you used. If only you had written at least a one-line comment to explain the technique.
The above situation is contrived, absolutely. The message is clear: document your code. Even if it is a one-line comment, use that single line to note the name of the (tricky or clever) technique the code is implementing.
Exercises
Exercise 1. Software documentation is more than just commenting your programming code. You need to know the type of documentation you are reading or writing. Is it a tutorial, an exposition, a how-to guide, or a reference manual? Learn more about different types of documentation.
Exercise 2. Writing software documentation is difficult. But do not despair. The Write the Docs project has a beginner’s guide to help you get started. Browse through the guide for tips on how to begin documenting your (first) project.
Exercise 3. The GHC project uses a tool called Haddock to extract documentation from annotated Haskell source files. Learn more about Haddock. In particular, you might want to browse the section on how to annotate and markup your Haskell source code.
Exercise 4. Let’s see how the function print
is documented. Load the page for the API documentation of print
. Locate a link called Source
. Click on the link to open the page that contains the corresponding source code of print
. Take a moment to examine how the source code of print
is annotated and documented.
Martin Fowler. Refactoring: Improving the Design of Existing Code, 2nd edition. Pearson Education, 2019, p.10. ↩
C, C++, Java, and JavaScript all use the delimiters
/*
and*/
to mark the beginning and end, respectively, of a multiline comment. Python uses the triple single quotation marks'''
(or the triple double quotation marks"""
) to delimit the start and end of a multiline comment. ↩