Developer Guide

The following details outline our standards and guidelines to producing high-quality code and documentation for Winter. While you are free to choose your own standards and guidelines for your own projects, the maintainer team of Winter recommends that you follow these standards for contributions either to the Marketplace, or to the core Winter CMS repositories.

PHP coding standards

Winter follows the PSR-2 Coding Style Guide as set forth by the PHP Framework Interop Group. This guide provides a great starting point for creating code that is easily understandable and familiar to developers. We intend to stick closely to the PSR-12 Extended Coding Style Guide in the future as well, but this is not a hard requirement at the present time.

Exceptions to the standard

Due to historical choices and technical limitations, we have some exceptions to these guidelines in Winter.

Controller methods can have a single underscore

The PSR-2 guidelines state that methods must be in camelCase format. However, in Backend controllers in Winter, AJAX handlers can be created using a suffix notation if they are connected to a "main" action. For example:

public function index()
    // This is the index page (index action)

public function index_onDoSomething()
    // AJAX handler only works on the index action

public function onDoSomethingElse()
    // AJAX handler works globally for all actions

An exception must be granted for this scenario. In other scenarios, underscores should be avoided in method names.

Curly braces for condition blocks

Historical code in Winter has previously used a format where all closing curly braces are on their own line, such as with the following:

if ($expr1) {
    // if body
elseif ($expr2) {
    // elseif body
else {
    // else body;

The PSR-2 and PSR-12 guidelines do not have any specific requirement for the placement of curly braces for condition blocks, but most examples on these guidelines, along with examples in other software and framework, try to keep the curly braces in the same line as the condition statement (unless it is multi-line).

if ($expr1) {
    // if body
} elseif ($expr2) {
    // elseif body
} else {
    // else body;

if (
    $expr1 &&
    $longExpression &&
) {
    // if body
} elseif (
    $seriously ||
    $thisIsSuperLongAndContrived ||
) {
    // elseif body
} else {
    // else body;

We intend to follow this format going forward. In addition, some historical code opted to not use curly braces at all for single line condition blocks.

if ($condition)

We do not recommend this format, and instead recommend all condition blocks use curly braces, even if they are single line.

if ($condition) {

Use of trailing commas

Although not specified one way or another in the PSR-2 and PSR-12 guidelines, Winter CMS highly recommends the use of trailing commas in multi-line arrays, especially for localization files. Trailing commas make it easier to perform maintenance on multiline array items without causing unnecessary visual clutter in diffs or version control history.


$items = [

Not recommended:

$items = [

Developer standards and patterns

Vendor naming

The vendor or author code in a namespace must begin with an uppercase character and should not contain underscores or dashes. These are examples of valid names:


These are examples of names that are not valid:

Repository naming

When publishing work to a repository, such as Git, use the following naming as a convention. Plugins should be named with a -plugin suffix and optional wn- prefix.


Themes should be named with the -theme suffix and optional wn- prefix.


Projects can be named with a wn- prefix and suffix indicating the project type (i.e. -site, -app, etc), although this is just a suggested convention and isn't enforced anywhere.

PHP Variable naming

Use camelCase everywhere except for the following:

  1. Database attributes and relationships should use snake_case
  2. Postback parameters and HTML elements should use snake_case
  3. Language keys should use snake_case

PHP Class naming

Use PascalCase for all classes. Underscores are to be avoided as they have been known to cause issues with autoloading due to the historical use of underscores in class names for pseudo namespacing before namespacing was officially supported in PHP. Numbers can be used if necessary (generally to denote that a class interacts with a specific version of a service, i.e. Conman vs Conman4) but should never be used at the start of a class name. Avoid all special characters.

If you find yourself wanting to use underscores in your class names for organizational purposes, don't. Use namespacing instead.


  • jet
  • myJet
  • my_jet
  • API_Jet


  • Jet
  • MyJet
  • API\Jet

HTML element naming

Form element names should use snake_case (underscores)

<input name="first_name">

Where the name is an array, the array keys can be either StudlyCase or snake_case.

<input name="ForumMember[first_name]">
<input name="forum_member[first_name]">

Element IDs should be camel case or hyphen-case (dashes)

<div id="firstNameGroup">
    <input id="firstName">

<div id="first-name-group">
    <input id="first-name">

Element classes names should use hyphen-case (dashes)

<div class="form-group">
    <input class="form-control">

View file naming

Partial views should begin with an underscore character. Whereas Controller and Layout views do not begin with an underscore character. Since views are often found in a single folder, the underscore (_) and dash (-) characters can be used to organise the files. A dash is used as a substitute for a space character. An underscore is used as a substitute for a slash character (folder or namespace).

index_fancy-layout.htm       <== Index\Fancy layout
form-with-sidebar.htm        <== Form with sidebar
_field-container.htm         <== Field container (partial)
_field_baloon-selector.htm   <== Field\Baloon Selector (partial)

View files must end with the .htm file extension.

Class naming

Classes commonly are placed in the classes directory. There is a number of class suffixes and prefixes that we recommend to use.

  1. Manager
  2. Builder
  3. Writer
  4. Reader
  5. Handler
  6. Container
  7. Protocol
  8. Target
  9. Converter
  10. Controller
  11. View
  12. Factory
  13. Entity
  14. Engine
  15. Bag

Don't get naming paralysis. Yes, names are very important but they're not important enough to waste huge amounts of time on. If you can't think up a good name in five minutes, move on.

Event naming

When specifying event names. The term after is not used in Events, only the term before is used. For example:

  1. beforeSetAttribute - this event comes before any default logic.
  2. setAttribute - this event comes after any default logic.

NOTE: This is true for the vast majority of cases, however the events present for default model events like boot, create, delete, fetch, restore, save, update, & validate all have before & after variants to match the model method events.

Where possible events should cover global and local versions. Global events should be prefixed with the module or plugin name. For example:

// For global events, it is prefixed with the module or plugin code

// For local events, the prefix is not required

Avoid using terms such as onSomething in event names since the word bind/fire represent this action word.

It is good practise to always pass the calling object as the first parameter to the global event, the local event should not need this. Local events take priority over global events when halting, or come first when processing.

// Local event
if ($this->fireEvent('beforeAddContent', [$message, $view], true) === false) {

// Global event
if (Event::fire('mailer.beforeAddContent', [$this, $message, $view], true) === false) {

When expecting multiple results, it is easy to combine the arrays like so:

// Combine local and global event results
$eventResults = array_merge(
    $this->fireEvent('form.beforeRefresh', [$saveData]),
    Event::fire('backend.form.beforeRefresh', [$this, $saveData])

Database table naming

Tables names should be prefixed with the author and plugin name.


Boolean column names should be prefixed with is_


This is because the model attributes can conflict, for example, public $visible; in the Model class conflicts with a database column with the same name. Some column names are exceptions, for example notify_user.

If your plugin extends tables belonging to other plugins, the added column names should be prefixed with the author and plugin name:


The author and plugin name acronym is also acceptable as a prefix:


Component naming

Component classes are commonly place in the components directory. The name of a component should represent its primary function.

To display a list of records, use the List suffix, eg:


To display a single record, use the Details suffix, eg:


Using the suffix helps avoid conflicts with controller and model names. Alternatively you can name components without the suffix, for cases when the name is descriptive and does not conflict:


Controller naming

Controllers are commonly are placed in controllers directory, for backend controllers. The name of a controller should be in plural form, for example:


Model naming

Models are commonly are placed in models directory. The name of a model should be in singular form, for example:


When dynamically extending other plugin's models, you should prefix the field with at least the plugin name. This helps to avoid potential future conflicts if that plugin is updated to add new relationships that could conflict with your dynamic relationships.

User::extend(function($model) {
    $model->hasOne['forum_member'] = ['Winter\Forum\Models\Member'];

The fully qualified plugin name is also acceptable, for example:


Model scopes

Model scopes should always return the scoped QueryBuilder instance to support scope chaining. If a scope is not returning a QueryBuilder instance then it is not a scope and should be a regular / static method instead.

// Valid scope method
public function scopeWithValidUser(Builder $query, User $user)
    return $query->where('user_id', $user->id);

// Invalid scope method
public function scopeWithValidUser(Builder $query, User $user)
    return $query->where('user_id', $user->id)->get();

// Valid standalone method returning a collection
public function getValidUser(User $user)
    return $this->withValidUser($user)->get();

Scope naming can be tricky as it's important to avoid conflicts with other aspects of the model's API, such as relationships & attributes. Generally speaking, scopes that receive additional parameters should be prefixed with apply to indicate they are being applied to the query. Defined as:

public function scopeApplyUser(Builder $query, User $user)
    return $query->where('user_id', $user->id);

Then applied to the model as:


Whilst apply is the ideal prefix name for those situations, here are some other prefixes we recommended for chained scopes:

  • is
  • for
  • with
  • without
  • filter

Class guidance

These points are to be considered in a relaxed fashion:

  1. In classes, properties and methods should be declared as protected in favor of private so that all classes can be used as base classes. Similarily, static::someMethodOrConstantOrProperty is preferred to self::someMethodOrConstantOrProperty, again to make it easier to extend classes.
  2. If a property contains a single value (not an array), make the property public instead of a get/set approach.
  3. If a property contains a collection (is an array), make the property protected with get getProperties, getProperty and setProperty.

Writing documentation

Your contributions to the Winter documentation are very welcome. We ask that you follow these guidelines when committing or proposing changes to the documentation.

Grammar & language

The Winter documentation is written in GitHub Flavored Markdown and should adhere to this specification.

At the moment, the Winter documentation only supports the English language. Documentation should be provided in US English. Where possible, you should strive to use natural language that would sound like a person verbally speaking to you, however, the maintainer team are more than happy to assist you with suggestions to this.

A number of our users use translation tools such as Google Translate to read our documentation, so where possible, you should avoid using too much jargon that may not be easily translateable.


Each page should have a single # Title (<h1>) at the top of the page that provides the title for the entire page. For all other headings, you should follow a logical heading structure, with main sections using a ## Secondary Heading (<h2>) title and sub-sections using a ### Tertiary Heading (<h3>) title.

You do not need to provide anchors for the headings, as these will be automatically generated.

# The main title

This is some introductory content for the page itself.

## Section heading

The section may have some introductory content here.

### Sub-section heading

This is the content of the first sub-section.

### Sub-section 2 heading

This is the content of the second sub-section.

## Next section heading

Table of contents

The Winter documentation site automatically generates a table of contents based on the heading structure of the page. You do not need to manually provide a table of contents. The table of contents is generated from the section and sub-section headings (<h2> and <h3>).


You may use bold (**bold**) and italicised (*italicised*) emphasis as required, but you should strive to use it sparingly. Avoid using either emphasis type for variable or path names and opt to use inline code blocks instead, as this will prevent these names from being translated by translation services such as Google Translate and ensure that people who use these tools will still be able to find these paths or variables.


It is encouraged to use linking as much as possible throughout the documentation, in order to provide users the most context possible during browsing the documentation. However, we recommend that you do not add links to the same article twice in the same section or sub-section, unless it is to another anchor point on the page.

External links are encouraged, but please confirm that the external resource is useful and operational. External links will always be opened up in a new tab.


We recommend the use of inline code code blocks for all path, variable and single line code references in the documentation. This provides a visual clue that the reference relates to code or the Winter files, and as stated above, will prevent the references from being translated. Inline code blocks can be created by wrapping the reference with a single backtick (`) at the start and end of the reference.

For larger, multiple line code blocks, we recommend the use of the code block format, by adding three backticks (```) to the start and end of the code block. You may also specify the language of the code immediately after the first set of backticks to enable the code highlighting feature of code blocks.

This is a code block for the Markdown language.

**It should highlight parts of the language.**

Will be converted into:

This is a code block for the Markdown language.

**It should highlight parts of the language.**

File structures

File structures can be demonstrated in Winter documentation with the treeview language inside a code block. The tree view language is a special language which uses a tree diagram to traverse the file and folder hierarchy.

A sample of a file structure can be found below:

|-- subfolder1/
|-- subfolder2/
|   |-- image.jpeg
|   |-- photo.png
|   `-- document.pdf
|-- subfolder3/
|   |-- sub2folder/
|   |    `-- sub3file    # A comment about this file
|   `-- sub2folder2/
|-- index.htm
`-- .hidden_file        # This file is hidden and should be slightly transparent

Which is then converted to the following code block:

|-- subfolder1/
|-- subfolder2/
|   |-- image.jpeg
|   |-- photo.png
|   `-- document.pdf
|-- subfolder3/
|   |-- sub2folder/
|   |    `-- sub3file    # A comment about this file
|   `-- sub2folder2/
|-- index.htm
`-- .hidden_file        # This file is hidden and should be slightly transparent

To break down this structure:

  • A folder is a name followed by a / character.
  • A file is a name with no / character suffixed.
  • A |-- tag indicates a child item of the item above. The pipe symbol (|) should align with the first character of the parent item. A space should always proceed the tag.
  • A `-- tag is the same as the |-- tag but indicates the last child of the item.
  • A | (pipe) should be use for parent folders that are still "open" while their child folders and files are being traversed. You must proceed a pipe with 3 spaces if there is any content to follow.
  • A # comment can be left at the end of any line. It cannot be used within the tree structure.

This feature also supports the output of the tree command-line utility which is available on most OS systems, allowing you to create the file structure in your OS and print a similar diagram to the one above. Use tree -Fa --dirsfirst <path> to print the friendly path for item, include hidden files and list directories first.

> tree -Fa --dirsfirst folder/

# folder/
# ├── .hidden_file
# ├── index.htm
# ├── subfolder1/
# ├── subfolder2/
# │   ├── document.pdf
# │   ├── image.jpeg
# │   └── photo.png
# └── subfolder3/
#     └── sub2folder/
#         └── sub3file

Which can then be converted to the following:

├── subfolder1/
├── subfolder2/
│   ├── document.pdf
│   ├── image.jpeg
│   └── photo.png
├── subfolder3/
│   └── sub2folder/
│       └── sub3file
├── .hidden_file
└── index.htm

NOTE: The tree command may print out indented lines using a character that looks to be a space character, but is not. If this is the case, you may need to add the --charset=ascii option to the command, which will print a diagram similar to the first example.

Copyright © 2023 Winter CMS
Edit on GitHub