- Code Style Guide - Creedengo
This document defines the coding standards for source files in the plugin. It is primarily intended for the plugin team but can be used as a reference by contributors.
The structure of this document is based on the Google Java Style reference and Springframework Code Style and is a work in progress.
Source files must be encoded using UTF-8
.
In Progresss
A source file consists of the following, in this exact order:
- Package statement
- Import statements
- Exactly one top-level class
Exactly one blank line separates each of the above sections.
In Progresss
The following governs how the elements of a source file are organized:
- static fields
- normal fields
- constructors
- (private) methods called from constructors
- static factory methods
- JavaBean properties (i.e., getters and setters)
- method implementations coming from interfaces
- private or protected templates that get called from method implementations coming from interfaces
- other methods
equals
,hashCode
, andtoString
Note that private or protected methods called from method implementations should be placed immediately below the methods where they're used. In other words if there 3 interface method implementations with 3 private methods (one used from each), then the order of methods should include 1 interface and 1 private method in sequence, not 3 interface and then 3 private methods at the bottom.
Above all, the organization of the code should feel natural.
Braces mostly follow the Kernighan and Ritchie style (a.k.a., "Egyptian brackets") for nonempty blocks and block-like constructs:
- No line break before the opening brace but prefixed by a single space
- Line break after the opening brace
- Line break before the closing brace
- Line break after the closing brace if that brace terminates a statement or the body of a method, constructor, or named class
- Line break before else, catch and finally statements
Example:
return new MyClass() {
@Override
public void method() {
if (condition()) {
something();
}
else {
try {
alternative();
}
catch (ProblemException ex) {
recover();
}
}
}
};
90 characters is the preferred line length we aim for. In some cases the preferred length can be achieved by refactoring code slightly. In other cases it's just not possible.
90 is not a hard limit. Lines between 90-105 are perfectly acceptable in many cases where it aids readability and where wrapping has the opposite effect of reducing readability. This is a judgement call and it's also important to seek consistency. Many times you can learn by looking how specific situations are handled in other parts of the code.
Lines between 105-120 are allowed but discouraged and should be few.
No lines should exceed 120 characters.
The one big exception to the above line wrapping rules is Javadoc where we aim to wrap around 80 characters for maximum readability in all kinds of contexts, e.g. reading on Github, on your phone, etc.
When wrapping a lengthy expression, 90 characters is the length at which we aim to wrap. Put the separator symbols at the end of the line rather on the next line (comma separated arguments, etc). For instance:
if (thisLengthyMethodCall(param1, param2) && anotherCheck() && yetAnotherCheck()) {
// ....
}
Add two blank lines before the following elements:
static {}
block- Fields
- Constructors
- Inner classes
Add one blank line after a method signature that is multiline, i.e.
@Override
protected Object invoke(FooBarOperationContext context, AnotherSuperLongName name) {
// code here
}
For inner-classes, extra blank lines around fields and constructors are typically not added as the inner class is already separated by 2 lines, unless the inner class is more substantial in which case the 2 extra lines could still help with readability.
Try as much as possible to put the implements
, extends
section of a class declaration on the same line as the class itself.
Order the classes so that the most important comes first.
Constant names use CONSTANT_CASE
: all uppercase letters, with words separated by underscores.
Every constant is a static final
field, but not all static final
fields are constants. Constant case should therefore be chosen only if the field is really a constant.
Example:
// Constants
private static final Object NULL_HOLDER = new NullHolder();
public static final int DEFAULT_PORT = -1;
// Not constants
private static final ThreadLocal<Executor> executorHolder = new ThreadLocal<Executor>();
private static final Set<String> internalAnnotationAttributes = new HashSet<String>();
Avoid using single characters as variable names. For instance prefer Method method
to Method m
.
- A file should look like it was crafted by a single author, not like a history of changes
- Don't artificially spread things out that belong together
Choose wisely where to add a new setter method; it should not be simply added at the end of the list. Perhaps the setter is related to another setter or relates to a group. In that case it should be placed near related methods.
- Setter order should reflect order of importance, not historical order
- Ordering of fields and setters should be consistent
Wrap the ternary operator within parentheses, i.e. return (foo != null ? foo : "default");
Also make sure that the not null condition comes first.
Use the org.springframework.util.Assert.notNull
static method to check that a method argument is not null
. Format the exception message so that the name of the parameter comes first with its first character capitalized, followed by "must not be null". For instance
public void handle(Event event) {
Assert.notNull(event, "Event must not be null");
//...
}
Always add @Override
on methods overriding or implementing a method declared in a super type.
A class that is only a collection of static utility methods must be named with a Utils
suffix, must have a private
default constructor, and must be abstract
. Making the class abstract
and providing a private
default constructor prevent anyone from instantiating it. For example:
public abstract MyUtils {
private MyUtils() {
/* prevent instantiation */
}
// static utility methods
}
A field of a class should always be referenced using this
. A method of class, however, should never be referenced using this
.
The use of var
for variable declarations (local variable type inference) is not permitted. Instead, declare variables using the concrete type or interface (where applicable).
From experience with issue #42, when an exception occurs during Sonar analysis with maven command, the consequence is a crash. If this phase is launched in a CI/CD pipeline, this crash will block the pipeline. Thus, our plugins will be deleted until next corrections.
Please, make a maximum of controls in the code to avoid ClassCast exceptions and thus, avoid uninstalling our plugin from SonarQube.
The following template summarizes a typical use for the Javadoc of a method.
/**
* Parse the specified {@link Element} and register the resulting
* {@link BeanDefinition BeanDefinition(s)}.
* <p>Implementations must return the primary {@link BeanDefinition} that results
* from the parsing if they will ever be used in a nested fashion (for example as
* an inner tag in a {@code <property/>} tag). Implementations may return
* {@code null} if they will <strong>not</strong> be used in a nested fashion.
* @param element the element that is to be parsed into one or more {@link BeanDefinition BeanDefinitions}
* @param parserContext the object encapsulating the current state of the parsing process;
* provides access to a {@link org.springframework.beans.factory.support.BeanDefinitionRegistry}
* @return the primary {@link BeanDefinition}
*/
BeanDefinition parse(Element element, ParserContext parserContext);
In particular, please note:
- Use an imperative style (i.e. Return and not Returns) for the first sentence.
- No blank lines between the description and the parameter descriptions.
- If the description is defined with multiple paragraphs, start each of them with
<p>
. - If a parameter description needs to be wrapped, do not indent subsequent lines (see
parserContext
).
The Javadoc of a class has some extra rules that are illustrated by the sample below:
/*
* Interface used by the {@link DefaultBeanDefinitionDocumentReader} to handle custom,
* top-level (directly under {@code <beans/>}) tags.
*
* <p>Implementations are free to turn the metadata in the custom tag into as many
* {@link BeanDefinition BeanDefinitions} as required.
*
* <p>The parser locates a {@link BeanDefinitionParser} from the associated
* {@link NamespaceHandler} for the namespace in which the custom tag resides.
*
* @author Rob Harrop
* @since 2.0
* @see NamespaceHandler
* @see AbstractBeanDefinitionParser
*/
- The order of tags for class-level Javadoc is:
@author
,@since
,@param
,@see
,@deprecated
. - The order of tags for method-level Javadoc is:
@param
,@return
,@throws
,@since
,@see
,@deprecated
. - In contrast to method-level Javadoc, the paragraphs of a class description are separated by blank lines.
The following are additional general rules to apply when writing Javadoc:
- Use
{@code}
to wrap code statements or values such asnull
. - If a type is only referenced by a
{@link}
element, use the fully qualified name in order to avoid an unnecessaryimport
declaration.
Tests must be written using JUnit Jupiter (a.k.a., JUnit 5).
The only exceptions to the above rule are test classes in the spring-test
module that specifically test Spring's integration with JUnit 4 and TestNG.
Each test class name must end with a Tests
suffix.
Use AssertJ for assertions.
Use the BDD Mockito support.
Unit test are mandatory. If you can, use Test Driven Development method to write your tests. Check coverage of your tests with SoanrQube integration of our code.