MA
MATLAB Architecture Patterns
MATLAB code organization patterns for large-scale scientific computing and toolbox development
Details
Language / Topic
MATLABCategory
Architecture
Rules
balanced
- Organize related functions into packages using `+packageName` folders and call them as `packageName.functionName()` to avoid name collisions.
- Use MATLAB classes (`classdef`) for domain objects with state and behavior — value classes for data containers, handle classes for objects that need shared mutable state.
- Separate data loading, computation, and visualization into distinct functions — scripts that mix all three are impossible to test and reuse.
- Use `handle` class inheritance to implement observer patterns: define abstract `notify` events and let dependent objects subscribe with `addlistener`.
- Keep toolbox code in `src/`, live scripts for exploration in `notebooks/`, unit tests in `tests/`, and compiled MEX binaries in `mex/` — separate concerns by folder.
- Use `inputParser` or `arguments` blocks for keyword-style arguments in complex functions rather than positional argument lists that callers must memorize.
- Version control `.m` files and `.slx` Simulink models with Git; use `slxdiff` or model comparison tools to review model changes in code review.