@cloudglides/nox 1.1.6 → 2.0.0

package/IMPROVEMENTS.md

@@ -1,58 +0,0 @@
-# Improvements Made (Hour 2)
-
-## Batch Operations
-- Added `batch(count, fn)` - Execute custom function n times
-- Added `floats(count)` - Generate array of floats
-- Added `ints(count, max)` - Generate array of integers
-- Added `bools(count, probability)` - Generate array of booleans
-
-## RNG API Enhancements
-- Added `range(min, max, step)` - Pick random value from stepped range
-- Added `choice(arr)` - Pick random element from array
-- Better error messages with TypeError vs RangeError distinctions
-
-## Statistical Improvements
-- Added `kolmogorovSmirnovTest(data)` - KS goodness-of-fit test
-- Added `meanTest(data, expected)` - Test mean of distribution
-- Added `varianceTest(data, expected)` - Test variance of distribution
-- All tests return detailed result objects with pass/fail indicators
-
-## Sampling Enhancements
-- Improved `weightedPick()` with comprehensive validation
-- Improved `weightedSample()` with type checking
-- Improved `reservoirSample()` with better error handling
-
-## Generator Optimizations
-- Removed modulo bias in `nextInt()` - uses rejection sampling instead
-- Applied to: PCG64, Xorshift64, Splitmix64, MT19937
-- Ensures uniform distribution across all ranges
-
-## Performance & Caching
-- Added crypto cache in entropy source for performance
-- Added `clearCryptoCache()` export for testing
-- Better entropy mixing with proper BigInt operations
-
-## Type Safety
-- Added `IGenerator` interface for type consistency
-- Added `GeneratorConstructor` type definition
-- Added generic types to shuffle, pick, sample functions
-- Added test result interfaces: `TestResult`, `KSTestResult`
-- Complete TypeScript definitions for all new APIs
-
-## Documentation
-- Updated README with batch operations examples
-- Added weighted sampling section
-- Added statistical tests section
-- Enhanced generator selection documentation
-
-## Testing
-- Added comprehensive test suite (test/comprehensive.js)
-- Added advanced features test (test/advanced.js)
-- All tests pass successfully
-
-## Code Quality
-- Total lines: 1107 → 1318 (+211)
-- Files modified: 10
-- Lines added/changed: 286
-- Consistent error handling patterns
-- Full backward compatibility maintained

package/PERFORMANCE.md

@@ -1,69 +0,0 @@
-# Performance Benchmarks
-
-Benchmarks run on Node.js v20.19.6 on modern hardware.
-
-## RNG Operations (per second)
-
-| Operation | Rate |
-|-----------|------|
-| `nextFloat()` | 2.5M |
-| `nextInt(100)` | 2.8M |
-| `int(1, 100)` | 1.8M |
-| `bool()` | 3.5M |
-| `choice()` | 1.5M |
-
-## Batch Operations
-
-| Operation | Rate |
-|-----------|------|
-| `floats(1000)` | 2.5k/sec |
-| `ints(1000, 100)` | 2.2k/sec |
-| `bools(1000)` | 2.8k/sec |
-
-## Distributions
-
-| Distribution | Rate |
-|--------------|------|
-| `normal()` | 1.0M |
-| `exponential()` | 1.5M |
-| `uniform()` | 3.5M |
-| `poisson()` | 0.8M |
-
-## Sequence Operations
-
-| Operation | Rate |
-|-----------|------|
-| `shuffle(100)` | 14k/sec |
-| `sample(100, 50)` | 25k/sec |
-| `pick()` | 1.5M |
-
-## Generators
-
-All generators show similar performance characteristics:
-- **PCG64**: Fast, cryptographic quality
-- **Xorshift64**: Fastest, good distribution  
-- **Splitmix64**: Very fast, good avalanche
-- **MT19937**: Good period, slower output
-
-## Optimization History
-
-### v1.1.5
-- 17% faster `nextFloat()` (division → multiplication)
-- 53% faster `nextInt()` (fast path for small max)
-- Pre-allocated arrays in batch operations
-- Removed unnecessary array copies in sampling
-
-### v1.1.4
-- Fixed unbiased distribution in `nextInt()`
-- Improved variance calculations
-
-### v1.1.3
-- Browser compatibility fixes
-
-## Tips for Best Performance
-
-1. **Reuse RNG instance**: Creating new RNG is cheaper than recreating distributions
-2. **Batch operations**: Use `.floats(n)` instead of loop with `.nextFloat()`
-3. **Small ranges**: `nextInt()` fast-paths for `max < 65536`
-4. **Avoid shuffling large arrays**: Use reservoir sampling or weighted sampling instead
-5. **Cache distribution parameters**: Pre-compute lambda, mean, stddev if used repeatedly