This repository has the notes of my learning and thinking in how to design and implement a software system in a systematic method. I focus on software systems because I am a software developer, not having too much experience in dealing with hardware. However, because software runs on top of hardware, hardware will inevitably get involved into the discussion to some extent.
Why is this topic important, at least to me? My answer is: because I want to reproduce success in every piece of software development task. Based on my personal experience, I have realized that, without a systematic method or process, I oftentimes solve the tasks at hand on an ad hoc basis: I may solve it for this particular task well enough, but I'm not sure if I can solve the next task the same well enough or even better.
- Model the target software system as a state machine. Related:
- "Write-ahead logs (WAL)" or similar technologies might be important to implement correct state transition.
- Exception safety.
- Design the architecture.
- See Martin Fowler: Software Architecture Guide: Identify the "important stuff".
- Define the scope of the target software system in order to distinguish the errors that can be handled and the errors that can't be handled.
- Report and handle errors correctly. See the sub-folder Error-Handling.
- Information security.
- Performance (or the so-called "non-functional requirements").
- Operation: We can't simply look at the software system itself. We must consider its upstream (i.e., how to get the needed material in order to reconstruct the software system automatically) and downstream (i.e., once the software system is upgraded, how to deploy it to the field automatically).
Every Computer System is a State Machine (or see the discussion on HN).
TODOs:
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent