UML isn't Dead
If you’ve studied software design at the university, there is no chance you missed UML diagrams. But, “when did you last use UML diagrams?” and the answer unanimously came across as “UML is pretty much dead, I used sequence diagrams occasionally, and that’s pretty much it”. I did some research, and I found some interesting stuff. Buckle up and hold on as we go through this ride.
It all started when I had to generate class diagrams for a data ingestion engine(2020) I was working on. I quickly found a Python library and stitched it to our Pre-Commit, generating a new diagram per commit.
Later in 2022, when I wanted to write a Github action, I used the same blog post as a reference and the generated ERD images as a part of the code review comment.
Are there other tools in the same arena? Is it possible to generate full fletched diagrams just by using codebase as your source of truth? Are there tools like this available already? Like Simon Brown, I hate diagramming tools. Too much mouse involvement, never up to date.
UML diagrams are no longer used for system design. I’m not sure how people are designing systems these days. The code is extremely user-friendly. It’s easier to visualize things from the code base, or is it so?
A quick search on jobs from indeed and LinkedIn shows that the UML skills have been taken out of the Software Engineers’ plate and moved to Solution Architect and Business Analyst’s bucket.
What does this shift mean?
As systems become complex, you need a tool that can express the software specifics in an understandable format to both tech and non-tech stakeholders. This makes UML perfect for Veteran Softwarians and enterprise architects who stick with UML. It's a great fit, from brainstorming your system design to developing and documenting them. There are tools in place to help us.
UML is widely used in software engineering and recognized as a de facto standard for modeling software systems. It is often used in the early stages of software development to help stakeholders understand and communicate the design of a system. UML is used in various contexts, including business process modeling, enterprise architecture, and system engineering.
- Class Diagram.
- Component Diagram.
- Deployment Diagram.
- Object Diagram.
- Package Diagram.
- Profile Diagram.
- Composite Structure Diagram.
- Use Case Diagram.
- Activity Diagram.
- State Machine Diagram.
- Sequence Diagram.
- Communication Diagram.
- Interaction Overview Diagram.
Among these, the class and the sequence diagram are the most commonly used
ERD diagrams are used to model the logical structure of a database. They depict the relationships between entities (e.g., tables or views) in a database and the attributes that define them. ERD diagrams are used to design and document a database's structure, typically used in the early stages of database development.
The C4 model is a software architecture model used to describe and communicate the structure of a software system. Simon Brown developed it to improve the effectiveness of software architecture documentation.
The C4 model is based on four levels of abstraction: context, container, component, and code. Each level represents a different perspective on the system, and the model provides a way to visualize and communicate the relationships between these levels.
One key difference between the C4 model and UML is that the C4 model is specifically focused on software architecture. At the same time, UML is a broader modeling language used to design various systems.
As the exploration began at the start of the blog, the goal was to generate a diagram from source code. Below are some language-wise tools that create diagrams.
PlantUML is an open-source tool allowing users to create diagrams from a plain text language. Besides various UML diagrams, PlantUML supports various software development-related formats
Alice -> Bob: Authentication Request
Bob --> Alice: Authentication Response
Alice -> Bob: Another authentication Request
Alice <-- Bob: Another authentication Response
ZenUML is a tool similar to plantUML. It claims to have better DSL syntax than PlantUML. PlantUML still has the market because of the ecosystem and the support for all UML diagrams.
These tools have a structure that will let you write these diagrams in text format. We are skipping the general-purpose diagramming tools such as Draw.io.
Mermaid can create a variety of Charts and graphs, Gantt, Class, ERD, and Sequence diagrams. It has its own syntax to generate diagrams. You can try it live at https://mermaid.live/
Websequence Diagrams is a simple and easy-to-use online tool. It comes with an online editor with many pre-defined examples to customize. It supports PlantUML syntax.
DBDocs uses the DBDigram.io tool is specifically designed to generate ERD diagrams based on the DB schema. Similar to the websequence diagram, it has a code-based editor. They support up to 10 diagrams for free.
Dbdiagram.io uses DBML as our syntax for defining databases. DBML has open-sourced markup language used to describe database schema structure. The team behind DBdiagram also develops DBML.
Eraser is a beautiful visual + code-based tool that lets you document your cloud architecture and ERD.
Draw.io might look like a drag-and-drop UI on the surface. But here is something that most people don’t know about it. It supports plantUML. You import a diagram using plantUML.
In the upcoming sections, we will be discussing language-specific diagram generators. Before we do that, It is important to pause here for a minute and appreciate the greatness of the GraphViz library. This library is the base for some of the famous visualization projects and is pretty much used in all the libraries we are going to discuss in the next section.
Graphviz is open-source graph visualization software. Graph visualization is a way of representing structural information as diagrams of abstract graphs and networks. The source of major diagramming tools, all the above tools work on top of this.
One downside of GraphViz is heavy(700MB). If you use any of the libraries in CI/CD, you are spinning up a 700MB+ container every time. Not very resource friendly
Prisma is an open-source data access layer that connects databases and manipulates data. It is built on the Node.js runtime and provides a simple and expressive API for working with databases.