From af277e21733948df122ffc210343c9b54cade7a9 Mon Sep 17 00:00:00 2001 From: nsfisis Date: Tue, 18 Aug 2026 01:57:02 +0900 Subject: test(process-executor): benchmark execute_async with criterion The `single` group compares the argv form against the string form, which goes through an extra `/bin/sh -c`; the `concurrent` group varies the job count around `max_jobs` so the semaphore throttle is visible as a throughput knee. `execute_async`'s futures are `!Send`, so they run on a current-thread runtime and overlap only through `join_all` within a single task. Co-Authored-By: Claude Opus 5 (1M context) --- Cargo.toml | 1 + 1 file changed, 1 insertion(+) (limited to 'Cargo.toml') diff --git a/Cargo.toml b/Cargo.toml index 574e800a..657f7da6 100644 --- a/Cargo.toml +++ b/Cargo.toml @@ -30,6 +30,7 @@ async-trait = "0.1.89" base64 = "0.22.1" bzip2 = "0.6.1" chrono = { version = "0.4.44", features = ["serde"] } +criterion = { version = "0.8.2", features = ["async_tokio", "html_reports"] } fastrand = "2.4.1" flate2 = "1.1.9" futures = "0.3.32" -- cgit v1.3.1-4-g156e