It is an interface for a task. The toolkit's ports are designed to work on their own. For example:
you can use the
http_server module without importing the
templates one, and the other way around
(taking only the dependencies you need for your application).
They are implementations of a functionality (Port) for a given product/technology. Clients should only use ports' code (not Adapters specific code), this makes it easy to switch among different adapters with minimum impact.
Adapters are independent of each other, but you can use several adapters for the same port in a single application.
Singleton object to manage a cross toolkit aspect. I.e.: Serialization, Injection or Settings.
The project is composed of modules, each module provides a single functionality. There are three kinds of modules:
- The ones that provide functionality that does not depend on different implementations, like
hexagon_scheduler or hexagon_core. Their name always starts with the
hexagon_prefix. These modules can depend on several Ports, but never on Adapters (see below).
- Modules that define a "Port": these are interfaces to a feature that may have different
implementations (ie: port_http_server or port_store). They cannot be used by themselves and in
their place, an adapter implementing them should be added to the list of dependencies. These
modules' names start with the
port_prefix. Ports are independent of each other.
- Adapter modules, which are Port implementations for a given tool, store_mongodb, and messaging_rabbitmq are examples of this type of module. Adapter names must start with their port name.
The Hexagon Core module is used by all other libraries, so it would be added to your project anyway just by using any adapter.
Core utilities like, logging, serialization, and dependency injection. Toolkit's ports are designed to use core functionalities. You can use a third party DI library instead of using the Core one. It depends on Logback and Jackson.
The main features are the following:
- Helpers: JVM information, a logger and other useful utilities.
- Dependency Injection: bind classes to creation closures or instances and inject them.
- Instance Serialization: parse/serialize data in different formats to class instances.
The following libraries provide extra features not bound to different implementations. They will not use dependencies outside the Hexagon toolkit.
- Settings: load settings from different data sources and formats. This properties are made available to other components through the SettingsManager.
- Scheduling: this module allows services to execute tasks periodically using Cron expressions. However, you have to be careful to not run tasks twice if you have many instances.
- Web: this module is meant to ease web application development. Provides helpers for generating HTML and depends on the HTTP Server and Templates ports.
These modules define features that need a specific implementation. You can use many implementations of each port at the same time. You can even provide a custom implementation if you want to optimize a particular use case.
These are the implemented ports:
- HTTP Server: describes how to use HTTP routing and HTML templates for Web services.
- HTTP Client: documentation to use the HTTP client module to connect to other services.
- Storage: gives an overview of how to store data using different data stores.
- Messaging: how to support asynchronous communication with messages through message brokers.
- Templates: describes how to render pages using template engines.