Programming style is a set of rules or guidelines used when writing the source code for a computer program. It is often claimed that following a particular programming style will help programmers to read and understand source code conforming to the style, and help to avoid introducing errors.
A classic work on the subject was The Elements of Programming Style, written in the 1970s, and illustrated with examples from the Fortran and PL/I languages prevalent at the time.
The programming style used in a particular program may be derived from the Coding conventions of a company or other computing organization, as well as the preferences of the author of the code. Programming styles are often designed for a specific programming language (or language family): style considered good in C source code may not be appropriate for BASIC source code, and so on. However, some rules are commonly applied to many languages.
Contents |
Good style is a subjective matter, and is difficult to define. However, there are several elements common to a large number of programming styles. The issues usually considered as part of programming style include the layout of the source code, including indentation; the use of white space around operators and keywords; the capitalization or otherwise of keywords and variable names; the style and spelling of user-defined identifiers, such as function, procedure and variable names; and the use and style of comments.
Programming styles commonly deal with the visual appearance of source code, with the goal of requiring less human cognitive effort to extract information about the program. Software has long been available that formats source code automatically, leaving coders to concentrate on naming, logic, and higher techniques. As a practical point, using a computer to format source code saves time, and it is possible to then enforce company-wide standards without debates.
Indent styles assist in identifying control flow and blocks of code. In some programming languages indentation is used to delimit logical blocks of code; correct indentation in these cases is more than a matter of style. In other languages indentation and whitespace do not affect function, although logical and consistent indentation makes code more readable. Compare:
if (hours < 24 && minutes < 60 && seconds < 60) { return true; } else { return false; }
or
if (hours < 24 && minutes < 60 && seconds < 60) { return true; } else { return false; }
with something like
if ( hours< 24 && minutes< 60 && seconds< 60 ) {return true ;} else {return false ;}
The first two examples are probably much easier to read because they are indented in an established way (a "hanging paragraph" style). This indentation style is especially useful when dealing with multiple nested constructs.
The ModuLiq Zero Indent Style groups with carriage returns rather than indents. Compare all of the above to:
if (hours < 24 && minutes < 60 && seconds < 60) return true; else return false;
Python uses indentation to indicate control structures, so correct indentation is required. By doing this, the need for bracketing with curly braces (i.e. { and }) is eliminated. On the other hand copying and pasting Python code can lead to problems, because the indentation level of the pasted code may not be the same as the indentation level of the current line. Such reformatting can be tedious to do by hand, but some text editors and IDEs have features to do it automatically. There are also problems when Python code could be rendered unusable when posted on a forum or web page that removes whitespace, though this problem can be avoided where it is possible to enclose code in whitespace-preserving tags such as "<pre> ... </pre>" (for HTML), "[code]" ... "[/code]" (for bbcode), etc.
Haskell similarly has the off-side rule which lets indentation define blocks; however, unlike in Python, indentation is not compulsory in Haskell — curly braces and semicolons can be (and occasionally are) used instead.
It is often helpful to align similar elements vertically, to make typo-generated bugs more obvious. Compare:
$search = array('a', 'b', 'c', 'd', 'e'); $replacement = array('foo', 'bar', 'baz', 'quux'); // Another example: $value = 0; $anothervalue = 1; $yetanothervalue = 2;
with:
$search = array('a', 'b', 'c', 'd', 'e'); $replacement = array('foo', 'bar', 'baz', 'quux'); // Another example: $value = 0; $anothervalue = 1; $yetanothervalue = 2;
The latter example makes two things intuitively clear that were not clear in the former:
Arguments against vertical alignment are
In those situations where some whitespace is required the grammars of most free-format languages are unconcerned with the amount that appears. Style related to whitespace is commonly used to enhance readability. There are currently no known hard facts (conclusions from studies) about which of the whitespace styles are having the best readability.
For instance, compare the following syntactically equivalent examples of C code.
int i; for(i=0;i<10;++i){ printf("%d",i*i+i); }
versus
int i; for (i=0; i<10; ++i) { printf("%d", i*i+i); }
versus
int i; for (i = 0; i < 10; ++i) { printf("%d", i * i + i); }
versus
int i; for( i = 0; i < 10; ++i ) { printf( "%d", i * i + i ); }
The use of tabs to create white space presents particular issues when not enough care is taken because the location of the tabulation point can be different depending on the tools being used and even the preferences of the user.
As an example, one programmer prefers tab stops of four and has his toolset configured this way, and uses these to format his code.
int ix; // Index to scan array long sum; // Accumulator for sum
Another programmer prefers tab stops of eight, and his toolset is configured this way. When he examines his code, he may well find it difficult to read.
int ix; // Index to scan array long sum; // Accumulator for sum
One widely used solution to this issue may involve forbidding the use of tabs for alignment or rules on how tab stops must be set. Note that tabs work fine provided they are used consistently, restricted to logical indentation, and not used for alignment:
class MyClass { int foobar(int qux, // first parameter int quux); // second parameter int foobar2(int qux, // first parameter int quux, // second parameter int quuux); // third parameter };