wsr

Template migration and repository tooling

flowchart LR
    U[Template snapshot] --> R[Reviewed WSR adaptations]
    R --> V[Tooling verification]
    V --> P[Preserved recovery material]

WSR initially adopted a rendered snapshot of carlosferreyra/rust-template at 74d68f0, adopted on October 7, 2026. The WSR baseline was 468444aa7d49fc08cbda3c08a19604e2b1c8e8b7.

On October 8, repository tooling was synchronized with template commit 62488bc. This is a reviewed tooling merge; WSR’s product and design documents remain its own.

The migration retains sixteen product crates and their existing Rust sources, version 0.0.2, license, release history, and product documentation. It adds private tools/xtask automation, the Cargo alias, a stable Rust toolchain with rustfmt/Clippy, resolver 3, and product Rust 1.85 declarations. Product default members exclude xtask so ordinary Cargo commands retain their original scope. No workflow engine, provider, or sandbox is implemented by this migration.

Develop

cargo xtask check
cargo xtask test
cargo xtask build
cargo xtask ci
cargo xtask doctor

check runs formatting, locked all-feature/all-target compilation and warnings-denied Clippy. test also runs locked workspace tests; build then builds the release workspace. ci adds locked documentation with Rustdoc warnings denied while retaining caller-supplied Rustdoc flags. WSR adds cargo xtask corpus for offline native Python references and pinned upstream snapshot integrity. --list inventories cases, --case ID selects one, and --prepare fastapi|uv explicitly fetches verified bare Git source without checking out or executing workloads. This is a WSR-owned extension, not an imported template capability; preserve it during future tooling merges. Ordinary Cargo tests now include the corpus bridge and require Python 3.11+ and Git. See the corpus guide. Passing scaffold/native-reference checks does not verify WSR workflow execution or isolation.

Tooling inherits the template’s workspace lint policy. Product crates do not yet inherit that policy: the assessment identified 75 additional warnings, principally missing API documentation. Enabling it is a separate focused change. Product sources and CLI help remain the current scaffold.

Grow and update deliberately

Start behavior as a module, then extract a project-prefixed crate when it owns a coherent responsibility and a useful testable seam. Existing placeholder crates do not dictate future architecture. PLAN.md still owns those decisions.

cargo xtask add <semantic-name> --dry-run

New scaffolds support planning and ownership checks. Existing CLI sources, CI, contribution guides, and crate inventory remain WSR-owned; their scaffolds intentionally refuse conflicting content. Do not add generated markers just to bypass that protection or replace these documents with starter text. The template’s README, PLAN, CONTEXT, and maintainer CI are not WSR’s design.

Generated projects are snapshots. There is no automatic template synchronization. For updates, review changes against the pinned revision, merge selected tooling/assets in a branch, retain WSR-specific verification/release behavior, and update this record. Action references live in tools/xtask/assets/tooling.toml; optional tool versions are selected at sync time. The checked-in reusable workflows are deliberately hand-maintained. Match their checkout/setup references to the registry when updating them. The cargo-dist-generated release workflow is a separate owner.

The October 8 merge adopts workspace dependency inheritance for xtask and CLI scaffolding, using concrete generation-resolved requirements: cargo_metadata 0.23.1, Clap 4.6.7, and toml_edit 0.25.16. Private xtask now inherits the shared workspace version. Sync updates stable Rust with rustfmt/Clippy and installs current stable crates.io releases with each release’s locked dependencies. Discovery accepts working local or global tools without exact-version checks. Full CI scaffolds request unversioned cargo-deny/typos-cli through the install action. The retired template-maintainer CI asset was removed; generation hooks and their placeholders are not imported into WSR.

WSR retains its product Rust 1.85 baseline instead of adopting the freshly generated template’s 1.99.0 baseline. It also retains locked all-feature checks, warnings-denied documentation, publication exclusions, preparation’s --no-push, and the existing CI structure. Cargo-dist is a documented exception to unversioned sync: an existing dist-workspace.toml selects the matching upstream Git tag, currently 0.33.0, preventing a downgrade to an older crates.io release. Sync does not rewrite workspace dependency requirements, the product MSRV, or release CI.

October 8 verification passed: fresh template generation, WSR’s standard cargo xtask ci, product checks on Rust 1.85.0, real release planning, and disposable-fixture scaffold checks for a private crate, primary CLI, and full CI. The template’s tool regression harness verified two sync upgrades and local/global discovery; WSR-specific probes verified the distribution pin, no-push preparation, private publication exclusions, and guarded changelog behavior without uploads. Hashes confirmed all 80 protected product/document/history files were unchanged, and the preserved stash list was identical. Hosted CI has not run for the synchronization edits.

CI and releases

The existing CI entrypoint, reusable workflows, check names, manual dispatch, credential handling, and release workflow are retained. Lint/tests invoke xtask. Checkout and Rust setup use the imported registry references. Local and hosted checks preserve locked dependencies, all-feature linting, and warnings-denied documentation. No scheduled audit was added during the initial migration.

The October 9 follow-up adds weekly Cargo/GitHub Actions Dependabot updates and individual cargo xtask ci --job format|lint|docs|test operations. Hosted jobs reuse these operations without repeating formatting or linting; aggregate local commands keep their existing checks. Private xtask explicitly requires Rust 1.86 while product crates retain Rust 1.85. Dist discovery now rejects executables that differ from the configured 0.33.0 version, retaining the Git-tag installer. The release changelog hook skips both generation and staging during dry runs. The release guide pushes the distribution tag separately instead of batching workspace tags with --follow-tags. The generated release workflow remains unchanged; actionlint records five narrow cargo-dist 0.33.0 ShellCheck exceptions, to be reviewed on generator upgrades. scripts/test-tooling.py exercises CI job isolation, configured dist selection/installation, and dry-run/execute hook behavior in temporary fixtures without installs or uploads.

Optional tools are installed only on request:

cargo xtask tools test
cargo xtask tools coverage
cargo xtask tools release

Executables install under ignored .xtask/tools; coverage setup also installs Rustup’s llvm-tools-preview component. Do not treat these development helpers as workload isolation.

The migration initially retained cargo-dist 0.30.3. The October 7 tooling refresh updated the registry and dist-workspace.toml to 0.33.0 and regenerates the release workflow with that version. Cargo-dist was installed from its matching upstream Git tag with --locked; other tools initially used exact crates.io versions, superseded by the October 8 sync behavior above. The refresh also updated typos to 1.51.1, install-action to 2.87.26, and the reusable workflows’ rust-cache pin to 2.9.2, and added Rustup self-updates to those workflows. Local Rustup 1.29.1 and stable Rust 1.99.0 were already current. The release workflow uses supported github-action-commits overrides for checkout 7.0.1, upload-artifact 7.0.2, download-artifact 8.0.2, and rust-cache 2.9.2, matching the registry. Regenerate it with .xtask/tools/bin/dist generate --mode ci after updating these settings; verify it with the same command plus --check. Existing release.toml, cliff.toml, CHANGELOG, and target list are preserved. release init remains a guarded generator and refuses the existing user-owned release configuration; this repository is already initialized.

The refresh was verified locally with cargo xtask tools all, cargo xtask doctor, cargo xtask ci, cargo xtask release plan, and the release workflow generation check. All seven optional tools then resolved to their pinned project-local installations; cargo-generate 0.25.0 was already available globally. Hosted CI was not run for those local edits.

cargo xtask release prepare
cargo xtask release plan

These require the available tools. Preparation defaults to cargo-release’s dry run. With explicit --execute, the imported wrapper still passes both --no-publish and WSR’s added --no-push; it may change versions, changelog, commits, and tags locally. Use a clean reviewable branch and handle publication separately. No release command was executed during the migration.

After migration, WSR added a separate publisher:

cargo xtask release publish
cargo xtask release publish --execute

It delegates to cargo-release’s workspace publish phase, defaults to a dry run, and skips versions already on the registry. Registry errors, including retry times, are returned without automatic waiting. Rerun the publisher to resume; preparation would bump versions again. Publication does not push commits or tags. See the release procedure.

The CLI and its fourteen supporting crates were published at version 0.0.3, and a fresh registry installation was verified. Internal dependencies retain versioned declarations. wsr-bench is an internal benchmark placeholder and is now private, alongside xtask; its workspace version can still advance without an upload. Published scaffolds remain scaffolds: registry availability does not establish runtime or security support.

Preservation and recovery

Before migration, all original Markdown documents and LICENSE were archived and checked by SHA-256. A committed-tree archive and verified Git bundle preserve the baseline history and original stash. Recovery locations and verification results are recorded outside the repository in the migration backup’s PROGRESS.md; backups are not committed as product files.

The original stash remains daee0885832df55a9a872037d3ce6741b9e6e72a. Its prototype code and older crates/xtask tooling were not applied. Restore only selected product work in an isolated recovery checkout after reconciling it with the architecture. Keep old root manifests, automation, and documents from overwriting the template migration or current decisions.

WSR keeps technical decisions/status in its existing documents; the organization’s .github repository continues to summarize that direction. Template adoption changes development tooling, not the accepted runtime design or its implementation status.

October 21 template sync

Merged tooling changes from local rust-template commit 7810ac016fc79a62fbcfaf1d297b493c1ca4b044 (October 17). This commit is on the local template branch; template origin/main still points to 62488bc. Canonical commands are now add <semantic-name>, init <capability>, and tools <group>; hidden legacy commands remain supported. New crate and CLI scaffolds generate intentionally failing unit, integration, documentation (for libraries), and executable test obligations, as appropriate. Existing product sources and tests remain intact. Agent scaffolding now creates the bundled CLI-first guide and refuses to overwrite user-owned instructions.

The merge retains WSR’s individual CI jobs, dist compatibility checks, locked checks, publication exclusions, no-push preparation, and dry-run changelog guards. Generation hooks and initial template crate placeholders are not imported.

Verification passed: cargo xtask ci, python3 scripts/test-tooling.py, python3 scripts/test-scaffolds.py, canonical help and add dry-run inspection, and git diff --check. Disposable library, binary, primary CLI, and companion CLI fixtures compiled their generated tests, failed with the expected obligations, and passed after completing the examples. Hosted CI has not run for this merge.