CakeDC Blog

TIPS, INSIGHTS AND THE LATEST FROM THE EXPERTS BEHIND CAKEPHP

CakeDC Git Workflow - An Introduction

Its been almost a year now since we released, and then later open sourced, the CakeDC Git Workflow at CakeFest 2013 in San Francisco. Since then, we've had loads of feedback, and have also experienced ourselves how it's revolutionized the way we work on projects.

When we first set out to define the workflow we had some issues which we wanted to resolve. The main ones being broken staging servers due to unstable branches, an unorganized planning of QA on a build, repeated efforts when testing code which is constantly changing, and messy repositories with no clear organization.

Having these problems at hand, we wanted to accomplish a couple of goals:

  • Maintain a master branch which is reliable as a stable and versioned code base
  • Provide a staged code base that's stable and best represents the upcoming version
  • Allow new releases to be comprised of multiple milestones (or sprints)
  • Allow developers to create features from the code developed by others
  • Allow the next milestone to start while the QA process is still active on the previous
  • Allow QA to review code on an isolated branch without affecting the stage server
  • Isolate bug fixing on separate branches to avoid active development during QA
  • Provide a process which can be planned around and scheduled for QA and releases

So, we set out to define a process which would allow us to meet these goals, and help us deliver projects, without the pain of the managing that process itself.

Organize and coordinate

When working with a team of managers, developers and testers, it becomes very important to keep your sanity by organizing and coordinating efforts on projects. When these projects are large in size and scope, that can become a difficult task, especially if you don't have a clearly defined process at hand. And that doesn't just mean defining a series of steps to follow, but a process which sets the team's direction, and facilitates the desired results.

The CakeDC Git Workflow does just that, by setting out a clear path to follow, and key points in which members of the team are involved, from managers and developers, through to QA testers and client review. These break down as the following:

  • Development: After gathering requirements and planning out a milestone this is the first phase. During this time the code base is actively worked on, and can be considered unstable, in a bleeding edge state. Each ticket is developed on a feature branched from the develop branch. Peer review would take place on each feature branch before it reaches develop.
  • QA: Once the first phase of development is complete the QA process begins. This is performed on an isolated branch, so the next milestone could commence. The acceptance criteria defined from the requirements would be applied here. Any bugs found by the testers are fixed on an issue branched from the qa branch.
  • Review: Once testing has concluded and the code base is considered stable it's merged to the stage branch, and a milestone is tagged. The client or product manager would now review the results and provide feedback.
  • Release: Once the work completed in milestones constitutes a new version of the application the code from stage is merged to master, and a release is tagged.

Iterating through milestones

At the core of the workflow is the concept of milestone development. A milestone represents a deliverable, and is broken down into 3 phases: development, qa and staging. Each of these has a dedicated branch in the repository, which holds the work completed at each step of the process, and ensures that all work done on the project follows through these phases.

The milestone also helps organize the development team as well as the client (product owner), as the workflow keeps everyone in a cycle, which helps avoid feature creep and sets clear and coherent objectives and responsibilities at each point in the process.

Quality as the driving factor

At CakeDC our ultimate objective is to deliver the highest quality possible. This means that all members involved with a project need to provide the best possible to meet that common goal. We do it because we care about what we're building, and want the result to match our expectations as to what the "best" means in each case.

Our workflow keeps that philosophy in high regard, as its designed to protect the code base at all times from anything which doesn't meet the grade. Each phase acts as a barrier to avoid the master branch from being compromised.

Latest articles

A CakePHP Docker Development Environment

We sponsor a monthly CakePHP training session (register here https://training.cakephp.org ) where we cover different topics about the framework. One of our sessions, the "Getting Started with CakePHP 4" is aimed to help developers starting a new project quickly and following the best practices.   Our previous "recommended" quick setting for a CakePHP development environment was using a vagrant box. See details here:  https://www.cakedc.com/jorge_gonzalez/2018/01/17/using-a-vagrant-box-as-quick-environment-for-the-getting-started-with-cakephp-training-session. However, we've switched internally to use docker as our primary development environment and also we started using docker in our training sessions.   Here's a quick overview of a simple docker based development environment for CakePHP.  

1. Create a new CakePHP project skeleton using 

composer create-project cakephp/app myproject   A new folder "myproject" will be created with a CakePHP project skeleton inside. Go to this new directory and proceed.  

2. Create a new "docker-compose.yaml" file with the following contents

version: '3' services:   mysql8:     image: mysql:8     restart: always     container_name: mysql     environment:         MYSQL_ROOT_PASSWORD: root         MYSQL_DATABASE: my_app         MYSQL_USER: my_app         MYSQL_PASSWORD: secret     volumes:       - ./:/application     ports:       - '9306:3306'     cakephp:     image: webdevops/php-apache:8.0     container_name: cakephp     working_dir: /application/webroot     volumes:       - ./:/application     environment:       - WEB_DOCUMENT_ROOT=/application/webroot       - DATABASE_URL=mysql://my_app:secret@mysql/my_app     ports:       - "8099:80"
 

3. Run "docker-compose up"

You'll create 2 containers named mysql and cakephp -  check the docker-compose configuration to see default database and users created in the mysql container, and the same environment params passed to the cakephp container via DATABASE_URL to allow the cakephp container to connect with the mysql database.   NOTE: the ports exposed are 9306 for mysql and 8099 for cakephp webserver. You can list them using docker-compose ps.  

4. Access your database and cakephp shell

  • To access the database you can use the command:
mysql --port 9306 -umy_app -psecret my_app   To restore a database dump for example, you can use the command: curl -L https://raw.githubusercontent.com/CakeDC/cakephp4-getting-started-session/master/my_app.sql |mysql --port 9306 -umy_app -psecret my_app   You can also configure any database tool to access the database in: localhost:9306  
  • To access the cakephp environment and shell you can use the command:
docker exec -it --user application cakephp bash   You'll go to the webroot folder, so in order to run the cake shell you'll need to: cd .. bin/cake 
  Now you have a working environment to play with the training session contents.   In this previous article, we covered another approach to setting up a local docker environment: https://www.cakedc.com/rochamarcelo/2020/07/20/a-quick-cakephp-local-environment-with-docker    We hope to see you in our next training session! https://training.cakephp.org   

Updating Model Layer

One reason to migrate from CakePHP 2.x to newer versions, is the very powerful ORM system that was introduced in CakePHP 3.x.  

Improved ORM Objects

The CakePHP model layer in CakePHP 3.x uses the Data Mapper pattern. Model classes in CakePHP 3.x ORM are split into two separate objects. Entity represents a single row in the database and it is responsible for keeping record state. Table class provides access to a collection of database records and describe associations, and provides api to work with a database. One notable change is afterFind callback removal. In CakePHP 3.x, it is possible to use Entity level getters to provide calculated fields on entity level.  

Association Upgrade 

In CakePHP 2.x associations are defined as arrays properties like this:     public $belongsTo = [         'Profile' => [             'className' => 'Profile',             'foreignKey' => 'profile_id',         ]     ];   In CakePHP 3.x and 4.x associations are declared in the initialize method. This gives much more flexibility in association configuration.     public function initialize(): void     {         $this->belongsTo('Profile', [                 'className' => 'Profile',                 'foreignKey' => 'profile_id',         ]);              }   Or using setters syntax, it could be done this way: public function initialize(): void     {         $this->belongsTo('Profile')         ->setForeignKey('profile_id')     }  

Behavior Upgrade

In CakePHP 2.x, behaviors are initialized as arrays properties:     public $actsAs = [         'Sluggable' => [             'label' => 'name',         ],     ];   In CakePHP 3.x and 4.x,  behaviors are configured in the initialize method. This gives much more flexibility in configuration, as in params it's possible to pass anonymous functions.     public function initialize(): void     {         $this->addBehavior('Sluggable', [             'label' => 'name',         ]);              }  

Validation Upgrade

In CakePHP 2.x, behaviors are  initialized as arrays properties:     public $validation = [         'title' => [             'notBlank' => [                 'rule' => ['notBlank'],                 'allowEmpty' => false,                 'required' => true,                 'message' => 'Title is required'             ],         ],     ];   In CakePHP 3.x and 4.x, validation is defined in validationDefault method which builds validation rules.     public function validationDefault(Validator $validator): Validator     {         $validator             ->scalar('title')             ->requirePresence('title', 'create')             ->notEmptyString('title');           return $validator;     }   Additionally, CakePHP introduced the buildRules method, which is where  described foreign keys constraints, uniqueness, or business level rules.     public function buildRules(RulesChecker $rules): RulesChecker     {         $rules->add($rules->existsIn(['user_id'], 'Users'));         $rules->add($rules->isUnique(['username'], __('username should be unique')));                            return $rules;     }  

Finder Methods

In CakePHP 2.x, the custom finder method is called twice - before and after fetching data from the database, which is defined by the $state parameter. Parameter $query contains current query state, and in $results passed data returned from database.     protected function _findIndex($state, $query, $results = array()) {         if ($state == 'before') {             $query['contain'] = ['User'];         } else {             // ...         }     }       In CakePHP 3.x, custom finder method accepts query object and some options passed from client code and returns an updated query. This allows for combining multiple finder methods in the same call, and has better grained finder logic.     public function findIndex(Query $query, array $options): Query     {         return $query->contain(['Users']);     }   The afterFind method could be implemented with the Query::formatResults method, which accepts an anonymous function to map each collection item.

Why Database Compression?

Nowadays people are not concerned about how large their database is in terms of MB. Storage is cheap. Even getting cheap SSD storage is not a big deal.    However, this is true if we are talking about hundreds of MB or even several GB, but sometimes we get into a situation where we have massive amounts of data (i.e Several tables with lots of longtext columns). At this point it becomes a concern because we need to increase the hard disk size, and find ourselves checking to see  if the hard disk is full several times per day or week, etc.   Now, if you have faced a situation like this before, it's time to talk about database compression. Compression is a technique, developed theoretically back in the 1940s but actually implemented in the 1970s. For this post we will focus on MySQL compression, which is performed using the open-source ZLib library. This library implements the LZ77 dictionary-based compression algorithm.   Before going into MySQL compression details, lets name some of the main DBMS and their compression techniques:

  • MySQL: ZLib (LZ77) [1]
  • Oracle: Oracle Advanced Compression (Proprietary)[2]
  • Postgres: PGLZ or LZ4 (if added this option at compilation level) [3]
  • DB2: Fixed-length compression or Huffman in some systems [4]
  So, now that we know this useless information, lets learn how to implement this in MySQL.   Firstly, you need to know that you CAN'T enable compression if:
  • Your table lives into `system` tablespace, or
  • Your tablespace was created with the option `innodb_file_per_table` disabled.
  It is important to test if the compression is the best solution for you.  If you have a table with a lot of small columns, you will probably end up with a larger-size table after "compressing" because of the headers and compression information. Compression is always great when you have longtext columns which can be heavily compressed.   Then, to enable compression for a table, you just need to include the following option when your table is created, or execute it as part of an alter statement: ROW_FORMAT=COMPRESSED These are the basics but you may find more useful information in MySQL manual.   You can also take a look at Percona which implements a Column level compression. This is interesting if you have a table with a lot of small fields and one large column, or if you have to optimize your database as much as possible. [6]   Finally, just say that even that storage is cheaper than ever, the amount of information has increased as well and we are now using and processing an incredible amount of data... so it looks like compression will always be a requirement.   I hope you find this information useful and please let me know if you have any questions or suggestions below in the comments section.

  [1]:https://dev.mysql.com/doc/internals/en/zlib-directory.html  [2]:https://www.oracle.com/technetwork/database/options/compression/advanced-compression-wp-12c-1896128.pdf  [3]:https://www.postgresql.org/docs/devel/runtime-config-client.html  [4]:https://www.ibm.com/docs/en/db2-for-zos/12?topic=performance-compressing-your-data  [6]:https://www.percona.com/doc/percona-server/8.0/flexibility/compressed_columns.html

We Bake with CakePHP