L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Développement · SKILL.md

run-rust-tests

run-rust-tests

Run Rust tests after making changes to verify correctness

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: run-rust-tests description: Run Rust tests after making changes to verify correctness

Rust Test Skill

Run Rust tests after making changes to verify correctness.

Arguments

  • No arguments: Analyze changes and run tests for affected crates only
  • all: Run all Rust tests
  • <crate>: Run tests for specific crate (e.g., /rust-test hyperloglog)
  • <crate> <test>: Run specific test in crate (e.g., /rust-test hyperloglog test_merge)

Arguments provided: $ARGUMENTS

Usage

Run this skill after modifying Rust code to ensure tests pass.

Instructions

  1. Check the arguments provided above:
    • If arguments are empty, determine affected crates:
      1. Check which files were modified in src/redisearch_rs/ using git status and git diff --name-only
      2. Map each modified file to its crate (the directory name directly under src/redisearch_rs/, e.g., src/redisearch_rs/hyperloglog/src/lib.rs → hyperloglog)
      3. Run tests for each affected crate:
        cd src/redisearch_rs && cargo nextest run -p <crate1> -p <crate2> ...
        
      4. If no Rust files were modified in src/redisearch_rs/, or if you cannot determine affected crates, run all tests
    • If all is provided, run all Rust tests:
      cd src/redisearch_rs && cargo nextest run
      
    • If a crate name is provided, run tests for that crate:
      cd src/redisearch_rs && cargo nextest run -p <crate_name>
      
    • If both crate and test name are provided, run the specific test:
      cd src/redisearch_rs && cargo nextest run -p <crate_name> <test_name>
      
  2. If tests fail:
    • Read the error output carefully
    • Fix the failing tests or the code causing failures
    • Re-run tests to verify the fix

Common Test Commands

# Test specific crate
cd src/redisearch_rs && cargo nextest run -p hyperloglog
cd src/redisearch_rs && cargo nextest run -p inverted_index
cd src/redisearch_rs && cargo nextest run -p trie_rs

# Run a specific test
cd src/redisearch_rs && cargo nextest run -p <crate_name> <test_name>

# Run tests under miri (for undefined behavior detection)
cd src/redisearch_rs && cargo +nightly miri test
Skills enregistrées