Any application (not just web application) has to be developed, tested and deployed (when approved by the customer). All of these steps are usually performed on different computers. You can imagine your application as a guest living in different hotel rooms all around the world. You want to create a nice compatible application which can easily live in different hotels without many problems. Because you (or somebody from your team) will be responsible for configuring and installing the application for each environment. Therefore you need to prepare your application for this and not hard-code any important setting into your source code.
What can be different:
If you can, try to have same versions of PHP and all libraries installed in every environment.
The environment for developers. Each developer has slightly different one. You should be very careful to design your application in a way that new developer does not have difficulties setting up the application when he joins your team. You will be responsible for explaining the setup to him so you will save your own time.
Usually one or few machines in your company where the application is installed before planned release and manually or automatically tested.
The environment used by your users/customers. You do not want to mess this up. Only tested application can be deployed here. Design and document the update process. Use automation scripts to avoid errors and mistakes (manual update is tedious). You should be able to revert to previous version quickly when update goes wrong.
Developing out-of-the-box software (like operating system, office applications package or video player) is much more difficult because you need to be prepared for thousands and thousands of different configurations. Custom software is usually controlled by company which created it and is installed only on limited number of machines with much more controlled configuration (concerning HW and OS).
Web applications are even better, they are installed on very few, usually only one, server.
Modern frameworks often use main configuration file called
.env. The idea is that you have a text file with
environment related key-value configuration pairs in the root of your project. This file should be ignored by Git to
avoid sharing your settings and passwords with other programmers (or their passwords with you). Anyway, Git repository
should contain an example file, e.g.
.env.example which you can copy and customize for current machine.
.env file should be hidden from visitors. You should let Apache web server see only the contents of public folder
to avoid opening URL such as
.env files in
# for Apache 2.2 and lower
Deny from all
# for Apache 2.4
Require all denied
There are libraries which implement parsing
.env file and reading values
from it. In your code you just use a function (e.g.
env()) which can read out a value from
.env file and return
Always have alternative/default value prepared for given environment variable variable.
.env file does not have to contain only database credentials. You can store many more configuration directives
that can activate or deactivate various functions of your application.
.env file is usually given and user of the application cannot do anything with predefined values.
Do you remember those two tables called
relation_type from your course project? They are typical
lookup tables. They are there to store possible types of something. The reason to use such lists is to allow certain
customisations for different instances of your application. When you install a new instance of your application,
you can add or remove values from these tables according to requirements of particular user. You can also translate
those values to a different language.
It is not a good idea to harcode such values into your source code. At least try to put the values into config file so they can be modified at once.
You can decide whether to put list values into config file or into the database depending on the relationship between
entities: if you can store the value into a text column and you do not plan to use the value much in SQL querries
(e.g. for filtering) of related record, you can put it into config file otherwise you better use lookup table with
foreign key (try to avoid
SELECT DISTINCT column FROM table).
The level of customisation is obviously limited by the design of application. There is a trade off between the level of customisation and complexity of source code. You should try to allow as much customisation as possible with relatively small complexity increase of source code to cover most user requirements.
It is a good practice to allow users of the application to modify values in lookup tables. The ability to modify these values can be conditionally dependent on user privileges (e.g. admin user).
You are developing an application, it will usually not be used only by you and not run on your computer. Keep this in mind.