Skip to content

Performance

Fakt’s compile-time impact and telemetry system.

Build Time Impact

Fakt uses intelligent caching across KMP targets:

First target compilation:

DISCOVERY: 1ms (100 interfaces, 21 classes)
GENERATION: 39ms (121 new fakes, avg 333µs/fake)
TOTAL: 40ms

Subsequent targets (cached):

compileKotlinJvm:     1ms (121 from cache)
compileKotlinAndroid: 1ms (121 from cache)
compileKotlinIosX64:  1ms (121 from cache)

Telemetry Configuration

Four log levels for debugging and performance analysis:

import com.rsicarelli.fakt.compiler.api.LogLevel

fakt {
    logLevel.set(LogLevel.INFO)   // Default: concise summary
    // LogLevel.QUIET  - Zero output (CI/CD)
    // LogLevel.DEBUG  - Detailed breakdown with FIR + IR details
}

Log Level Outputs

INFO (default):

✅ 10 fakes generated in 1.2s (6 cached)
   Discovery: 120ms | Analysis: 340ms | Generation: 580ms
   Cache hit rate: 40% (6/15)

DEBUG:

[DISCOVERY] 120ms - 15 interfaces, 3 classes
[FILTERING] 85ms - Cache hits: 6/15 (40%)
[ANALYSIS] 340ms
  ├─ UserRepository (18ms)
  ├─ Analytics (42ms)
  ├─ FIR + IR node inspection, type resolution

Includes full FIR + IR details, type resolution, etc. (~5-10ms overhead)

Cache Strategy

Fakt caches generated code across:

  • KMP targets (jvm, ios, android, etc.)
  • Incremental compilation runs
  • Clean builds (invalidates cache)

Best Practices

✅ Use QUIET in CI/CD

fakt {
    logLevel.set(LogLevel.QUIET)  // Zero overhead
}

✅ Use DEBUG for Troubleshooting

fakt {
    logLevel.set(LogLevel.DEBUG)  // ~5-10ms overhead
}

Next Steps