Class ParallelExecutor

Parallel executor implementation with dual execution modes:

Promise-based parallelism (default): File-level parallelism using Promise.race()
- Works with ts-morph Project instances (no serialization needed)
- Good for I/O-bound workloads
- 1.5-2x speedup on 100+ files
Worker thread parallelism: True multi-threaded execution
- Per-worker Project instances (isolated memory)
- Excellent for CPU-bound workloads
- 5-10x speedup on 500+ files
- Requires useWorkerThreads flag and tsConfigPath

Deprecated

The WorkerPool analysis path is deprecated since 1.1 and will be removed in 2.0; the MCP scans run in a scan worker instead.

Implements

ParallelExecutor

Index

Constructors

Methods

executeParallel getStats shutdown

Constructors

constructor

new ParallelExecutor(
    options?: {
        maxWorkers?: number;
        project?: Project;
        resourceManager?: ResourceManager;
    },
): ParallelExecutor
Parameters
- Optionaloptions: { maxWorkers?: number; project?: Project; resourceManager?: ResourceManager }
Returns ParallelExecutor

Methods

executeParallel

executeParallel(
    sourceFiles: SourceFile[],
    rules: OrchestratorAnalysisRule[],
    options?: ParallelOptions,
): Promise<AnalysisResult[]>
Execute analysis rules in parallel across files

Current implementation uses chunked batching with Promise.all for parallelism. This provides significant speedup while maintaining compatibility with ts-morph.
Parameters
- sourceFiles: SourceFile[]
  Files to analyze
- rules: OrchestratorAnalysisRule[]
  Rules to execute
- Optionaloptions: ParallelOptions
  Execution options
Returns Promise<AnalysisResult[]>
Analysis results

Implementation of IParallelExecutor.executeParallel

getStats

getStats(): WorkerPoolStats
Get worker pool statistics

Returns WorkerPoolStats
Implementation of IParallelExecutor.getStats

shutdown

shutdown(): Promise<void>
Shutdown worker pool gracefully

Returns Promise<void>
Implementation of IParallelExecutor.shutdown