
interface-segregation-principle
Use when designing interfaces. Use when implementing interfaces with methods you don't need. Use when forced to implement throw/no-op for interface methods.
Use when designing interfaces. Use when implementing interfaces with methods you don't need. Use when forced to implement throw/no-op for interface methods.
Interface Segregation Principle (ISP)
Overview
Clients should not be forced to depend on interfaces they don't use.
Many small, focused interfaces are better than one large "fat" interface. If an implementer must throw exceptions or provide no-ops for interface methods, the interface is too large.
When to Use
- Designing a new interface
- Implementing an interface with unused methods
- Forced to implement methods that don't apply
- Interface has more than 5-7 methods
- Different implementers use different subsets of methods
The Iron Rule
NEVER implement an interface method with throw or no-op.
No exceptions:
- Not for "it's what the interface requires"
- Not for "I'll provide both approaches"
- Not for "the caller can check capabilities"
- Not for "it's clearly documented as unsupported"
Providing both the violation and the correct approach is still providing a violation.
Detection: The "Throw/No-op" Smell
If your implementation looks like this, the interface is wrong:
// ❌ FAT INTERFACE
interface MultiFunctionDevice {
print(doc: string): void;
scan(): string;
fax(doc: string): void;
}
// ❌ VIOLATION: Forced to implement unusable methods
class BasicPrinter implements MultiFunctionDevice {
print(doc: string): void { /* works */ }
scan(): string { throw new Error("Not supported"); } // ← ISP violation
fax(doc: string): void { /* no-op */ } // ← ISP violation
}
The Correct Pattern: Segregated Interfaces
Split the fat interface into focused capabilities:
// ✅ CORRECT: Segregated interfaces
interface Printer {
print(doc: string): void;
}
interface Scanner {
scan(): string;
}
interface Fax {
fax(doc: string): void;
}
// Implement only what you support
class BasicPrinter implements Printer {
print(doc: string): void { /* works */ }
// No scan or fax - doesn't promise what it can't deliver
}
class AllInOne implements Printer, Scanner, Fax {
print(doc: string): void { /* works */ }
scan(): string { /* works */ }
fax(doc: string): void { /* works */ }
}
// Combined type for callers who need everything
type MultiFunctionDevice = Printer & Scanner & Fax;
Pressure Resistance Protocol
1. "The Interface Already Exists"
Pressure: "Implement this existing interface, handle unsupported methods"
Response: The interface is wrong. Propose splitting it.
Action:
"This interface forces implementers to provide throw/no-op for methods they don't support.
I recommend splitting into: [list focused interfaces].
Should I refactor the interface, or document this as tech debt?"
2. "Just Throw an Error"
Pressure: "Handle unsupported methods by throwing"
Response: Runtime errors for expected interface methods is a design failure.
Action: Split the interface so implementers only promise what they can deliver.
3. "I'll Provide Both Options"
Pressure: "Here's the violation you asked for AND here's the better way"
Response: Providing the violation at all enables bad code to ship.
Action: Provide ONLY the correct approach. Don't implement the fat interface.
4. "Callers Can Check First"
Pressure: "Add a supports(method) check"
Response: This is a workaround for bad design. Type system should enforce capabilities.
Action: Split interfaces so the type system does the checking at compile time.
Red Flags - STOP and Reconsider
If you notice ANY of these, the interface needs splitting:
- Implementing a method with
throw new Error - Implementing a method as no-op (empty body)
- Interface has 7+ methods
- Different implementers use different subsets
- Adding
supportsX()capability checks - Implementers have large blocks of unused methods
All of these mean: Split the interface.
Interface Design Guidelines
Size
- Ideal: 1-3 methods per interface
- Acceptable: 4-5 methods if highly cohesive
- Too large: 6+ methods - look for split opportunities
Cohesion Test
Ask: "Do ALL implementers need ALL these methods?"
- Yes → Keep together
- No → Split
Common Splits
| Fat Interface | Segregated Interfaces |
|---|---|
Repository<T> |
Readable<T>, Writable<T> |
Worker |
Workable, Eatable, Meetable |
MultiFunctionDevice |
Printer, Scanner, Fax |
FileSystem |
FileReader, FileWriter, FileDeleter |
UserService |
UserReader, UserWriter, UserAuth |
Quick Reference
| Symptom | Action |
|---|---|
| Method implemented as throw | Split interface |
| Method implemented as no-op | Split interface |
| 7+ methods in interface | Look for split |
supports() capability checks |
Split interface |
| Implementers ignore methods | Split interface |
Common Rationalizations (All Invalid)
| Excuse | Reality |
|---|---|
| "The interface already exists" | Interfaces can be refactored. |
| "Throwing makes it explicit" | Compile errors are better than runtime errors. |
| "I provided both approaches" | Providing the violation enables bad code. |
| "It's documented as unsupported" | Documentation doesn't fix design flaws. |
| "Many interfaces is complex" | Many small interfaces is simpler than one broken one. |
| "Callers can check capabilities" | Type system should do this, not runtime checks. |
The Bottom Line
No client should be forced to depend on methods it doesn't use.
When asked to implement a fat interface:
- Identify which methods are actually needed
- Propose segregated interfaces
- Implement only the focused interfaces
Never provide throw/no-op implementations. Never provide "both options." The fat interface is the problem - fix it.
You Might Also Like
Related Skills

coding-agent
Run Codex CLI, Claude Code, OpenCode, or Pi Coding Agent via background process for programmatic control.
openclaw
add-uint-support
Add unsigned integer (uint) type support to PyTorch operators by updating AT_DISPATCH macros. Use when adding support for uint16, uint32, uint64 types to operators, kernels, or when user mentions enabling unsigned types, barebones unsigned types, or uint support.
pytorch
at-dispatch-v2
Convert PyTorch AT_DISPATCH macros to AT_DISPATCH_V2 format in ATen C++ code. Use when porting AT_DISPATCH_ALL_TYPES_AND*, AT_DISPATCH_FLOATING_TYPES*, or other dispatch macros to the new v2 API. For ATen kernel files, CUDA kernels, and native operator implementations.
pytorch
skill-writer
Guide users through creating Agent Skills for Claude Code. Use when the user wants to create, write, author, or design a new Skill, or needs help with SKILL.md files, frontmatter, or skill structure.
pytorch
implementing-jsc-classes-cpp
Implements JavaScript classes in C++ using JavaScriptCore. Use when creating new JS classes with C++ bindings, prototypes, or constructors.
oven-sh
implementing-jsc-classes-zig
Creates JavaScript classes using Bun's Zig bindings generator (.classes.ts). Use when implementing new JS APIs in Zig with JSC integration.
oven-sh