Parallel Execution

Status

Implemented

Threading Model

• Single Bash instance: sequential (&mut self)
• Multiple Bash instances: parallel via tokio::spawn
• Filesystem: thread-safe via Arc<dyn FileSystem> + RwLock
• Arc::clone(&fs) shares one filesystem across instances; instances run in parallel sharing it

Benchmark

Run cargo bench --bench parallel_execution when changes touch:

• Arc, RwLock, shared state
• Interpreter, Bash, FileSystem
• Async paths (tokio::spawn, .await)
• Builtins (grep, awk, sed, etc.)

Key Metrics

Benchmark	What it measures
workload_types/*	Parallel vs sequential speedup
parallel_scaling/*	Scaling with session count (10–1000 sessions)
single_*	Individual operation overhead

Correctness at Scale

Throughput numbers are meaningless if sessions silently error out. The parallel_sessions_tests integration suite asserts that a 1000-session fan-out (each its own Bash, sharing one Arc<dyn FileSystem>) actually produces correct per-session output, and that concurrent sessions sharing a filesystem don't cross-contaminate. Run via just test (no extra features).

Expected Results

• Light workload: ~2x parallel speedup
• Medium workload: ~4x parallel speedup
• Heavy workload: ~7x parallel speedup

Must not degrade. Compare before/after.

See also

• Bashkit Architecture — shared-ownership decisions this builds on
• Builtin Commands — builtin execution under concurrency
• Performance Results — where benchmark results are kept
• Testing Strategy — concurrency test expectations