kimchi:simplicity-enforcement

Simplicity Enforcement

Overview

Complexity is the enemy. Every abstraction, configuration option, and "extensible" pattern is a cost. Pay it only when you must.

Core principle: Prefer simple solutions over clever ones. Always.

Violating the letter of these rules is violating the spirit of simplicity.

When This Applies

Every implementation decision. No exceptions.

The Iron Law

THE RIGHT AMOUNT OF COMPLEXITY IS THE MINIMUM NEEDED FOR THE CURRENT TASK

Three similar lines of code is better than a premature abstraction.

Principles

Prefer Duplication Over Wrong Abstraction

If you're not sure an abstraction is right:

• Duplicate the code
• Wait until you have 3+ similar cases
• Then extract with full understanding

```ruby # Duplication when pattern unclear class AvatarUploadService def upload(file) validate_size(file) validate_type(file) store_in_s3(file) end end

class DocumentUploadService def upload(file) validate_size(file) validate_type(file) store_in_s3(file) end end

Two cases. Let it sit. Extract when you see the third.
</Good>

<Bad>
```ruby
# Premature abstraction
class GenericUploadService
  def initialize(validator:, storage:, processor: nil)
    @validator = validator
    @storage = storage
    @processor = processor
  end

  def upload(file)
    @validator.validate(file)
    processed = @processor&.process(file) || file
    @storage.store(processed)
  end
end

Over-engineered for two use cases

YAGNI: You Aren't Gonna Need It

Don't build for hypothetical future requirements.

```ruby # Build what's needed now class ImageResizer def resize(file, dimensions) ImageProcessing::Vips .source(file) .resize_to_fill(*dimensions) .call end end ``` ```ruby # Hypothetical future support class ImageResizer STRATEGIES = { vips: VipsStrategy, imagemagick: ImageMagickStrategy, cloudinary: CloudinaryStrategy, # "in case we switch" }

def initialize(strategy: :vips) @strategy = STRATEGIES[strategy].new end end

</Bad>

### Hardcode First, Configure Later

If a value won't change soon, hardcode it.

<Good>
```ruby
AVATAR_SIZES = [32, 128, 512].freeze
BUCKET = 'avatars'

```ruby config.avatar_sizes = [32, 128, 512] # In YAML config config.avatar_bucket = ENV['AVATAR_BUCKET'] # When it's always 'avatars' ```

One Way to Do It

Don't provide multiple ways to accomplish the same thing.

```ruby user.avatar_url(:medium) ``` ```ruby user.avatar_url user.get_avatar_url user.fetch_avatar(size: :medium) Avatar.url_for(user) ```

Common Rationalizations

Excuse	Reality
"We might need this later"	Build it later. YAGNI.
"It's more flexible this way"	Flexibility you don't need is complexity you pay for now.
"DRY says extract it"	DRY applies at 3+ repetitions with clear pattern, not 2.
"Enterprise patterns are best practice"	Factory/Strategy/Observer for one use case is anti-practice.
"Configuration makes it reusable"	Configuration for fixed values is noise.
"What if requirements change?"	They will. Refactor then, with full context. Cheaper than guessing now.
"It's only a little more complex"	Complexity compounds. Every "little" addition adds up.

Red Flags — STOP and Simplify

• Writing a class with "Generic", "Base", or "Abstract" in the name for a single use
• Adding constructor parameters for "flexibility"
• Creating config files for values that won't change
• Writing a "factory" or "strategy" pattern
• Building plugin/extension points
• Adding feature flags for unrequested features
• Creating helpers/utilities for one-time operations
• Designing for hypothetical future requirements

ALL of these mean: STOP. Delete. Write the simple version.

Verification

Before completing:

• No abstractions without 3+ uses
• No configuration for fixed values
• No "pluggable" or "extensible" patterns for single use
• No multiple ways to do the same thing
• Could a junior developer understand this in 5 minutes?

Anti-Patterns

FORBIDDEN: "Enterprise" patterns for small features

Factory, Strategy, Observer, Visitor patterns are rarely needed. If you have one implementation, you don't need a pattern.

FORBIDDEN: Future-proofing

"What if we need to support X later?" — Build it later. You'll have more context then.

FORBIDDEN: DRY at all costs

Duplication is better than wrong abstraction. Two similar functions are fine. Extract at three.

FORBIDDEN: Backwards-compatibility shims

If something is unused, delete it. Don't rename to _var, re-export, or add # removed comments.