publish: move settings.toml into aube-settings; pin per-crate include lists by jdx · Pull Request #4 · endevco/aube

jdx · 2026-04-18T18:44:55Z

Summary

cargo publish --verify copies the crate to target/package/<name>-<ver>/ and re-runs the build there. aube-settings's build.rs was walking two .parent() hops up from CARGO_MANIFEST_DIR to find settings.toml at the workspace root — during publish verify those hops landed in target/, and settings.toml wasn't in the tarball anyway because cargo only packages files under the crate root.

Fix: move settings.toml into crates/aube-settings/ so it ships with the crate, and read it directly from CARGO_MANIFEST_DIR.

Also add an explicit include = [...] list to every crate's Cargo.toml. With implicit inclusion, anything tracked under a crate directory silently ends up in the published tarball. Explicit lists keep published crates minimal and prevent surprises:

crate	includes
`aube-settings`	`src/*/.rs`, `build.rs`, `settings.toml`
everything else	`src/*/.rs`

tests/ is intentionally excluded — the integration tests aren't useful to downstream consumers.

Verified locally

cargo build --workspace clean
cargo package -p aube-settings packages + verifies (10 files)
cargo package -p <each other crate> --no-verify produces the expected file list (6 files each, src + Cargo metadata only)
cargo run -p aube-settings --bin generate-settings-docs still generates docs/settings/*.md correctly

The other crates' --verify step can't run locally because they depend on aube-settings via crates.io and it's not published yet — release-plz handles this in production by publishing in dependency order.

Test plan

Next release-plz run publishes aube-manifest, aube-settings, and remaining crates to crates.io without the verify-build failure.

🤖 Generated with Claude Code

Note

Low Risk
Low risk: packaging/build-script path adjustments and documentation link updates, with no runtime behavior changes beyond making cargo publish --verify succeed.

Overview
Fixes aube-settings publishing by moving settings.toml under the crate and updating build.rs (and the settings docs generator) to read it from CARGO_MANIFEST_DIR, so cargo publish --verify can run from the packaged source.

Adds explicit include = [...] lists to each crate’s Cargo.toml to tightly control what ships in published tarballs (and updates generated docs/settings/*.md links to point at the new settings.toml location).

^{Reviewed by Cursor Bugbot for commit 2bbecd0. Bugbot is set up for automated code reviews on this repo. Configure here.}

… lists cargo publish --verify copies the crate into target/package/<name>-<ver>/ and builds from there. aube-settings' build.rs read settings.toml via two .parent() hops from CARGO_MANIFEST_DIR, landing at target/, and settings.toml wasn't in the tarball anyway (cargo only packages files under the crate root). Move settings.toml into crates/aube-settings/ so it ships with the crate, and read it directly from CARGO_MANIFEST_DIR. Also add an explicit include list to every crate so future accidentally tracked files (test corpora, generated docs, etc.) don't quietly end up in published tarballs. Current include lists: - aube-settings: src/**/*.rs, build.rs, settings.toml - everything else: src/**/*.rs tests/ is intentionally excluded — our integration tests aren't useful to downstream crate consumers.

greptile-apps · 2026-04-18T18:48:29Z

Greptile Summary

This PR fixes a cargo publish --verify failure for aube-settings by relocating settings.toml from the workspace root into crates/aube-settings/ and updating build.rs to read it via CARGO_MANIFEST_DIR rather than walking two parent directories up. It also adds explicit include = [...] lists to every crate's Cargo.toml to keep published tarballs minimal.

build.rs: settings_path now resolves from env!(\"CARGO_MANIFEST_DIR\") directly — fixes the root cause (the old .parent().parent() walk landed in target/package/ during verify, where settings.toml wasn't present)
crates/aube-settings/settings.toml: moved here from workspace root; content unchanged
crates/aube-settings/Cargo.toml: include list ensures the build script and its data file ship together
All other Cargo.toml files: include = [\"src/**/*.rs\"] — integration tests intentionally excluded
generate_settings_docs.rs: reads settings.toml correctly from the crate dir (line 32), but the hardcoded GitHub URL (line 182) still points to the old workspace-root location

Confidence Score: 4/5

Safe to merge with one small follow-up fix needed in the docs generator URL

The core fix is correct and well-reasoned — moving settings.toml into the crate and reading it via CARGO_MANIFEST_DIR is the right solution. The include lists across all crates are clean and consistent. The only concrete issue is a stale hardcoded GitHub URL in the docs generator that will produce broken links in every generated settings page; it's a one-line fix. The aube-cli [[test]] concern is low-impact.

crates/aube-settings/src/bin/generate_settings_docs.rs line 182 (stale URL); crates/aube-cli/Cargo.toml (excluded test target still declared)

Important Files Changed

Filename	Overview
crates/aube-settings/build.rs	Core fix: reads `settings.toml` from `CARGO_MANIFEST_DIR` directly instead of two `.parent()` hops; resolves `cargo publish --verify` breakage
crates/aube-settings/Cargo.toml	Adds explicit `include = ["src/*/.rs", "build.rs", "settings.toml"]` so the published tarball contains all files the build script needs
crates/aube-settings/settings.toml	File moved from workspace root into the crate directory so it ships with the published tarball; content unchanged
crates/aube-settings/src/bin/generate_settings_docs.rs	Updated to read `settings.toml` from `CARGO_MANIFEST_DIR`; stale GitHub URL on line 182 still points to old workspace-root location
crates/aube-settings/src/meta.rs	Doc comment updated to reflect `settings.toml` is now crate-local; no logic changes
crates/aube-cli/Cargo.toml	Adds `include = ["src/*/.rs"]`; the `[[test]]` declaration referencing `tests/e2e.rs` (not included) will cause `cargo test` to fail for anyone who downloads the crate
crates/aube-linker/Cargo.toml	Adds `include = ["src/*/.rs"]`; no other changes
crates/aube-manifest/Cargo.toml	Adds `include = ["src/*/.rs"]`; no other changes

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[cargo publish -p aube-settings] --> B[Package crate tarball]
    B --> C["Extract to target/package/aube-settings-ver/"]
    C --> D[cargo build --verify]
    D --> E[build.rs executes]
    E --> F["PathBuf::from(CARGO_MANIFEST_DIR).join('settings.toml')"]
    F --> G["Reads crates/aube-settings/settings.toml ✅"]
    G --> H["Generates settings_meta_data.rs + settings_resolved.rs → OUT_DIR"]
    H --> I[Build succeeds]

    subgraph OLD ["Before this PR (broken)"]
        X[build.rs] --> Y["CARGO_MANIFEST_DIR.parent().parent()"]
        Y --> Z["target/package/ — settings.toml not here ❌"]
    end

    subgraph NEW ["After this PR (fixed)"]
        F
        G
    end

Comments Outside Diff (1)

crates/aube-settings/src/bin/generate_settings_docs.rs, line 182 (link)

Stale GitHub link after settings.toml relocation

settings.toml was moved from the workspace root to crates/aube-settings/settings.toml in this PR, but the URL embedded in every generated docs page still points to the old workspace-root path. Clicking this link on any published docs page will land on a 404.

_{Reviews (2): Last reviewed commit: "publish: move settings.toml into aube-se..." | Re-trigger Greptile}

The 'generated from settings.toml' line in every docs/settings/*.md page is hard-coded in the generator. After moving settings.toml into crates/aube-settings/, the old workspace-root URL 404s. Update the template and regenerate all five pages.

Buildkite allocates a PTY for each job so tools emit colors. hk's spinner and mise's -v DEBUG stream both detect the TTY and flood the log — we counted 28,432 rows on build #4's `linux / test` step, most of them progress frames captured as newline-separated lines after Buildkite strips the cursor-control escapes. - mise.toml: pipe `hk check`/`hk fix` through `cat`, which breaks the PTY on both stdout and stderr and forces hk into its non-interactive path. ~10k spinner-frame rows per run go away. - .buildkite/bootstrap.sh: drop `-v` from `mise install`. `--locked` stays, INFO-level output stays; the ~15k DEBUG rows from the initial bootstrap go away. No wall-clock change — this is purely log hygiene so failures are findable and the Buildkite UI loads quickly.

## Summary - Pipe `hk check` / `hk fix` through `cat` in `mise.toml` so hk drops its spinner UI under Buildkite's PTY. Kills ~10k progress-frame rows per `linux / test` run. - Drop `-v` from `mise install` in `.buildkite/bootstrap.sh`. Kills ~15k DEBUG-level rows from bootstrap; `--locked` and INFO output remain. ## Why Buildkite agents run commands under a PTY so cargo/clippy emit colors. From hk and aube's perspective `isatty(stderr) == true`, so they render spinners. Buildkite's log capturer strips cursor-up/clear-line escapes but keeps the text, so each animation frame becomes its own log line. Concrete evidence from build #4's `linux / test` step: - Total log rows: **28,432** for a single step - `mise install -v` bootstrap alone: ~15,900 rows (DEBUG log flood) - `hk check --all --slow`: ~10,200 rows (spinner-frame flood) - Wall-clock breakdown unchanged — compile/lint time is ~3 minutes on a warm cache, it was just hard to see through the noise. Verified on warm-cache build #6 that the cache path itself is healthy (4.4G restored, zero `Compiling` messages in `cargo test`), so the 5-minute wall clock on the cold-start build #4 wasn't a cache bug — just a cold cargo target dir on pipeline init. ## Test plan - [ ] Next Buildkite run on this branch: `linux / test` log rows drop from ~28k to ~3k - [ ] `hk check --all --slow` still runs to completion and surfaces real lint errors (spinner gone, but the final error summary is what actually matters) - [ ] `mise install` still installs all tools; only the DEBUG chatter is gone 🤖 Generated with [Claude Code](https://claude.com/claude-code)  --- > [!NOTE] > **Medium Risk** > Changes how progress output is rendered in CI and modifies the Buildkite lint command pipeline; mistakes could hide failures (mitigated via `set -o pipefail`) or degrade progress UX. > > **Overview** > Reduces Buildkite log noise by making `aube` prefer CI-style (append-only) progress output when running in a detected CI environment even if stderr is a TTY (via new `is_ci` dependency). > > Buildkite’s `linux / test` step now runs `mise run lint` through `cat` with `pipefail` to prevent `hk` spinner frames from exploding log rows while still failing the step on lint errors, and `.buildkite/bootstrap.sh` drops verbose `mise install -v` to avoid DEBUG-level bootstrap spam. > > <sup>Reviewed by [Cursor Bugbot](https://cursor.com/bugbot) for commit 5cc91ed. Bugbot is set up for automated code reviews on this repo. Configure [here](https://www.cursor.com/dashboard/bugbot).</sup>

## Summary Real-world failure: [tuist/tuist#10584](tuist/tuist#10584) tried to swap pnpm for aube and CI hit `workspace package noora at noora has no \`version\` field` from [crates/aube/src/commands/install/mod.rs:1869](crates/aube/src/commands/install/mod.rs:1869). pnpm accepts the same repo without complaint. The unconditional check predates the resolver's current `workspace:*` / `workspace:^` / `workspace:~` / bare-`*` fallback. The comment's collision rationale ("two unversioned members under one dep_path") doesn't hold either: `dep_path` keys on `(name, version)` and two members have distinct names by construction. Fix: fall back to `"0.0.0"` when `version` is absent. `workspace:*` / `workspace:^` / `workspace:~` / bare-`*` siblings still link locally (those branches in `resolve_workspace` accept any ws version regardless). A specific range like `workspace:^2.0.0` against an unversioned member correctly fails to satisfy, matching pnpm. ## Test plan - [x] `cargo test` (full suite green) - [x] `cargo clippy --all-targets -- -D warnings` - [x] `cargo fmt --check` - [x] `mise run test:bats test/workspace.bats` — 19 pass, including the new regression #4 - [ ] CI green on push 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Skip empty bytes_segment so the phase-2 label doesn't render a doubled `· ·` separator on installs without `dist.unpackedSize` (Greptile P1). * Switch ETA to fetch-window throughput: snapshot completed at the set_phase("fetching") boundary, then divide remaining work by `fetch_elapsed_ms / (completed - baseline)`. Falls back to `ETA …` while the baseline or fetch window are missing. Eliminates the pessimistic "110s ETA after a 3s resolve" case (Greptile P2). * Reconcile estimated_bytes after filter_graph: track per-dep_path unpacked_size during streaming, then re-sum surviving dep_paths via the new InstallProgress::reconcile_estimated_bytes API. Mirrors the package-denominator reconcile applied next to it (Greptile P3). * Apply TARBALL_COMPRESSION_RATIO in TTY refresh_bytes_segment too — CI mode applied it; TTY was showing the raw unpacked sum (~3.3× inflated) (Cursor #1). * Refresh ETA on every inc_downloaded_bytes so it doesn't freeze during long downloads when no inc_total / inc_reused events fire. Documents the trade-off at FetchRow::drop where a refresh would require a larger refactor (Cursor #3). * Style only the ✓ green on the install / no-op summaries instead of the whole sentence; aligns with the CI heartbeat summary's styling so the success cue stays sharp without bleeding green across the line (Cursor #4 + user request). * Standardize the ✓ prefix on both Already-up-to-date paths (`print_install_summary` and `print_already_up_to_date`). Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

jdx enabled auto-merge (squash) April 18, 2026 18:46

jdx disabled auto-merge April 18, 2026 18:46

jdx merged commit 5d2fbc1 into main Apr 18, 2026
16 checks passed

jdx deleted the publish/crate-layout-fix branch April 18, 2026 19:01

jdx mentioned this pull request Apr 18, 2026

ci: quiet buildkite log noise from hk/mise progress output #5

Merged

3 tasks

This was referenced Apr 18, 2026

v1.0.0-beta.2: npm distribution and improved Linux compatibility #11

Merged

chore: release v1.0.0-beta.3 #22

Merged

jdx mentioned this pull request May 2, 2026

fix(install): allow workspace members without version field #480

Merged

5 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

publish: move settings.toml into aube-settings; pin per-crate include lists#4

publish: move settings.toml into aube-settings; pin per-crate include lists#4
jdx merged 2 commits intomainfrom
publish/crate-layout-fix

jdx commented Apr 18, 2026 •

edited by cursor Bot

Loading

Uh oh!

greptile-apps Bot commented Apr 18, 2026 •

edited

Loading

Comments Outside Diff (1)

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

jdx commented Apr 18, 2026 • edited by cursor Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Verified locally

Test plan

Uh oh!

greptile-apps Bot commented Apr 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Summary

Confidence Score: 4/5

Important Files Changed

Flowchart

Comments Outside Diff (1)

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

jdx commented Apr 18, 2026 •

edited by cursor Bot

Loading

greptile-apps Bot commented Apr 18, 2026 •

edited

Loading