Repository navigation
Slim agent guidance, add core lint, run rust-llm-tidy in verify #31
There are no files selected for viewing
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,109 @@ | ||
| # Migration to v1.4.2 | ||
|
|
||
| From `v1.4.1` to `v1.4.2`. | ||
|
|
||
| Slims the LLM guidance and moves `core`-over-`std` to a Clippy lint: | ||
|
|
||
| - Deletes the `.llm/{general,performance,documentation}.md` guidance files. | ||
| - Inlines the performance reminders into `src/AGENTS.md`. | ||
| - Runs [rust-llm-tidy] at the end of `.llm/verify.{sh,ps1}` when installed. | ||
| - Adds the `std_instead_of_core` Clippy lint to the workspace. | ||
|
|
||
| ## 1. Delete the guidance files | ||
|
|
||
| ```text | ||
| src/.llm/documentation.md | ||
| src/.llm/general.md | ||
| src/.llm/performance.md | ||
| ``` | ||
|
|
||
| ## 2. Replace `src/AGENTS.md` | ||
|
|
||
| ```markdown | ||
| After changes, find and run `.llm/verify.{sh,ps1}` to test + lint. | ||
| Print all output. | ||
|
|
||
| Performance: | ||
|
|
||
| - Prefer borrows: `&str` / `&[T]` returns, `&'static str` constants, | ||
| `Cow<'_, str>` for conditional ownership. | ||
| - `Box<str>` for immutable strings. | ||
| - Reuse buffers via `.clear()`. | ||
| - Const generics for compile-time branching such as | ||
| `<const LINE_NUMBERS: bool>`. | ||
| - Prefer performance-oriented crates such as `parking_lot` and `memchr`. | ||
| - Keep dependency footprint minimal. | ||
| ``` | ||
|
|
||
| ## 3. Add the tidy step to the verify scripts | ||
|
|
||
| In `.llm/verify.sh`, after the publish dry-run: | ||
|
|
||
| ```diff | ||
| echo "Publish dry-run..." | ||
| run_cmd cargo publish --dry-run --allow-dirty --quiet --workspace | ||
| + | ||
| +echo "Tidy..." | ||
| +if command -v rust-llm-tidy >/dev/null 2>&1; then | ||
| + run_cmd rust-llm-tidy | ||
| +else | ||
| + echo "Skipping; rust-llm-tidy not installed. Install: cargo install rust-llm-tidy-cli" | ||
| +fi | ||
|
|
||
| if [ "$EXIT_CODE" -eq 0 ]; then | ||
| ``` | ||
|
|
||
| In `.llm/verify.ps1`, after the publish dry-run: | ||
|
|
||
| ```diff | ||
| Write-Host "Publish dry-run..." | ||
| Invoke-LoggedCommand "cargo" @("publish", "--dry-run", "--allow-dirty", "--quiet", "--workspace") | ||
| + | ||
| + Write-Host "Tidy..." | ||
| + if (Get-Command rust-llm-tidy -ErrorAction SilentlyContinue) { | ||
| + Invoke-LoggedCommand "rust-llm-tidy" @() | ||
| + } else { | ||
| + Write-Host "Skipping; rust-llm-tidy not installed. Install: cargo install rust-llm-tidy-cli" | ||
| + } | ||
| } finally { | ||
| ``` | ||
|
|
||
| ## 4. Add the Clippy lint | ||
|
|
||
| Root `Cargo.toml`: | ||
|
|
||
| ```diff | ||
| [workspace] | ||
| resolver = "2" | ||
| members = ["src/my-project"] | ||
| + | ||
| +# Inherited by members via `[lints] workspace = true`. | ||
| +[workspace.lints.clippy] | ||
| +# Prefer `core` over `std` where both work. | ||
| +std_instead_of_core = "warn" | ||
| ``` | ||
|
|
||
| Every member crate (`src/my-project/Cargo.toml`, `src/cli/Cargo.toml` if present): | ||
|
|
||
| ```diff | ||
| # categories = ["command-line-utilities"] | ||
| + | ||
| +[lints] | ||
| +workspace = true | ||
| ``` | ||
|
|
||
| Then fix any hits, for example the benchmark: | ||
|
|
||
| ```diff | ||
| -use std::hint::black_box; | ||
| +use core::hint::black_box; | ||
| ``` | ||
|
|
||
| ## 5. Update the template version marker | ||
|
|
||
| ```diff | ||
| -reloaded-templates-rust:1.4.1 | ||
| +reloaded-templates-rust:1.4.2 | ||
| ``` | ||
|
|
||
| [rust-llm-tidy]: https://github.com/Sewer56/rust-llm-tidy |
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1 +1 @@ | ||
| reloaded-templates-rust:1.4.2 |
| Original file line number | Diff line number | Diff line change |
|---|---|---|
|
|
@@ -51,6 +51,13 @@ run_cmd cargo fmt --all --quiet | |
| echo "Publish dry-run..." | ||
| run_cmd cargo publish --dry-run --allow-dirty --quiet --workspace | ||
|
|
||
| echo "Tidy..." | ||
| if command -v rust-llm-tidy >/dev/null 2>&1; then | ||
| run_cmd rust-llm-tidy | ||
|
There was a problem hiding this comment. 🎯 Functional Correctness | 🟠 Major | ⚡ Quick win 🔎 Supported by static analysis🤖 get_repo_knowledge executed:
Length of output: 396 🏁 Script executed: #!/bin/bash
set -eu
printf '%s\n' '--- target files ---'
sed -n '45,65p' templates/general/src/.llm/verify.sh
sed -n '75,95p' templates/general/src/.llm/verify.ps1
sed -n '35,72p' docs/migration/v1.4.2.md
printf '%s\n' '--- tidy definitions and references ---'
rg -n --glob '!target' --glob '!node_modules' 'rust-llm-tidy|llm-tidy|dry-run|check' .Repository: Reloaded-Project/reloaded-templates-rust Length of output: 18213 🏁 Script executed: #!/bin/bash
set -eu
printf '%s\n' '--- rust-llm-tidy documentation in repository ---'
sed -n '1,110p' docs/migration/v1.4.0.md
printf '%s\n' '--- documentation conventions ---'
sed -n '1,90p' docs/AGENTS-DOCS.md
printf '%s\n' '--- repository status and relevant config ---'
sed -n '1,130p' templates/general/cargo-generate.toml
fd -i 'rust-llm-tidy*' . -t f -x sh -c 'echo "--- $1"; sed -n "1,140p" "$1"' sh {}Repository: Reloaded-Project/reloaded-templates-rust Length of output: 8926 🌐 Web query:
💡 Result: The rust-llm-tidy tool is an opinionated, work-in-progress Rust linter designed for LLM-assisted workflows, available via the Sewer56/rust-llm-tidy repository [1]. It provides a unified command-line interface to perform several code-tidying operations, including reordering source items, visibility narrowing, documentation linting, and markdown/doc-comment fixing [1]. Regarding your specific request for --validate and apply mode: --validate: This flag is used to verify the configuration without actually processing any files [1]. This is particularly useful for CI environments to ensure the.rust-llm-tidy.yml configuration is valid before running operations [1]. Apply Mode: When the tool is run (often via the rust-llm-tidy-action in CI/CD pipelines), it can operate in an apply mode [2][3]. In this mode, the tool automatically applies fixes to the target files in place [2]. For example, when used in a CI workflow on a pull request, apply mode is typically configured to commit and push these automated fixes back to the PR branch [2][3]. Configuration: The tool optionally uses a.rust-llm-tidy.yml configuration file, discovered by walking up from the current directory [1]. This file allows you to specify include/exclude rules or skip entire files [1]. If no configuration file is present, the tool uses default settings [1][3]. Citations:
🌐 Web query:
💡 Result: The rust-llm-tidy tool is a CLI utility written in Rust designed to lint and tidy codebases (specifically.rs and.md files) to improve compatibility or structure for LLM processing [1]. It provides specific flags for handling modifications and configuration validation [1]. Dry-Run Execution: To preview changes without applying them to your files, use the --dry-run flag [1]. For example: rust-llm-tidy --dry-run src/main.rs Configuration Validation: To verify the integrity of your configuration file (.rust-llm-tidy.yml) without processing any files, use the --validate flag [1]. This is particularly useful in CI environments to ensure the configuration is correctly formatted before running linting tasks [1]. rust-llm-tidy --validate Additional CLI usage details: - Default behavior: Running rust-llm-tidy without arguments processes files changed in the current git diff [1]. - Path specification: You can specify files or directories (e.g., rust-llm-tidy src) to process them recursively [1]. - Rule Selection: You can use --include and --exclude flags to run specific operations (e.g., fix, reorder, vis, lints) or skip them for a single invocation [1]. - Configuration Discovery: The tool automatically searches for a.rust-llm-tidy.yml file by walking up the directory tree to the repository root [1]. This can be overridden using the --config or --no-config flags [1]. Citations: Use non-mutating tidy verification. Bare 📍 Affects 3 files
🤖 Prompt for AI Agents |
||
| else | ||
| echo "Skipping; rust-llm-tidy not installed. Install: cargo install rust-llm-tidy-cli" | ||
| fi | ||
|
|
||
| if [ "$EXIT_CODE" -eq 0 ]; then | ||
| echo "All checks passed!" | ||
| else | ||
|
|
||
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,2 +1,13 @@ | ||
| After changes, find and run `.llm/verify.{sh,ps1}` to test + lint. | ||
| Print all output. | ||
|
|
||
| Performance: | ||
|
|
||
| - Prefer borrows: `&str` / `&[T]` returns, `&'static str` constants, | ||
| `Cow<'_, str>` for conditional ownership. | ||
| - `Box<str>` for immutable strings. | ||
| - Reuse buffers via `.clear()`. | ||
| - Const generics for compile-time branching such as | ||
| `<const LINE_NUMBERS: bool>`. | ||
| - Prefer performance-oriented crates such as `parking_lot` and `memchr`. | ||
| - Keep dependency footprint minimal. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
Check each deleted guidance path independently.
The current condition matches only the literal
.llm/{general,performance,documentation}.md. It does not detect individual references such as.llm/general.md. A generatedsrc/AGENTS.mdcan therefore contain a broken reference and still pass validation. Check each deleted path separately, while retaining the existing combined-form check.Proposed fix
📝 Committable suggestion
🤖 Prompt for AI Agents