ci(carrier): compile and test the Tauri host, which no job touched - #3541
Conversation
|
Warning Review limit reached
Next review available in: 21 minutes You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
📝 WalkthroughWalkthroughThe TypeScript suite now tests the generated Rust projection and Carrier Tauri host. CI installs required Linux packages, caches Cargo data, and allows more execution time. Carrier documentation now reflects this test coverage. ChangesCarrier Tauri CI coverage
Estimated code review effort: 3 (Moderate) | ~20 minutes Possibly related issues
Possibly related PRs
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Nothing in CI ran cargo for `apps/carrier/src-tauri`. Two consequences: the host's own protocol contract tests never executed, and the generated-contract path dependency added by the language-agnostic contracts work was never compiled, so a break in the Rust host's consumption of that projection could not be caught. The inert workflow whose path globs point at a directory that does not exist in this repository is an intentional migration stub, and is left alone — it is documented as such rather than being accidental coverage. The host's cargo test now runs in the existing carrier/hull job, whose path filter already covers the relevant trees. `TAURI_CONFIG` suppresses only the bundle-time external sidecar; this compiles and tests the host without packaging or launching the desktop app. A cargo cache is included. Without it every run recompiles the webkit bindings from scratch, and a slow required check is how people learn to ignore CI. The stale comment claiming CI does not test this crate is replaced with the actual coverage boundary. Proven by mutation: pointing the host at a generated constant that does not exist fails compilation in the new step, so the step really does compile the host's consumption of the projection. Honest limitation: GitHub Actions itself was not exercised. The commands were verified locally on macOS; the Ubuntu runner path — the apt prerequisites in particular — is unverified until this runs. Watch the first execution rather than assuming it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
5fec358 to
d875c35
Compare
|
thought (non-blocking): Accessibility audit (advisory)The sharded axe audit is report-only while the baseline and runtime budget mature.
Shard 1 reportShard 2 reportShard 3 report |
…uild Two one-liners from the review, both protecting a required check. The 15-minute ceiling was sized for a Node-only job. A cold Rust build of the Tauri host measured 3m33s wall and about 935 CPU-seconds on a 16-core Mac, and the Linux runner additionally compiles the webkit and gtk binding crates on fewer, slower cores. That build now shares the budget with apt, pnpm install, two dist builds, four lints, four vitest suites, the Rust projection test and the renderer production build. The compounding part is why this is not a nit: a timeout CANCELS the job, and a cancelled job skips the cache save, so the next run is cold again. An undersized ceiling on a required check is self-perpetuating. The apt step is the only one in any workflow here and had no retry, so a single transient mirror failure would block merge on a PR that merely touched the contracts package. Three attempts with a short backoff. The review refuted my own worry about the cargo cache masking a compile break, with evidence: both break-mutations went red against a warm target holding 319 cached dependency archives, because cargo recompiled only the changed crate. Restored caches carry archived mtimes older than the sources checkout writes, so the failure direction is over-rebuild, never under-rebuild. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 3
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In @.github/workflows/ts-suites.yml:
- Around line 91-94: Update the apt-get retry loop in the workflow so exhausting
all three attempts exits non-zero instead of succeeding because sleep is the
final command. Preserve the five-second delay between attempts, but explicitly
fail after the final unsuccessful apt-get update before continuing to package
installation.
- Around line 94-99: Update the prerequisite installation command in the
workflow step to include apt-get’s non-interactive option while preserving the
existing package list and no-recommends behavior, so the Tauri dependencies
install without prompting before pnpm ts-suites runs.
In `@scripts/ts-suites.mjs`:
- Around line 141-146: Update the `run` invocation for `test Carrier Tauri Rust
host` to include Cargo’s `--locked` flag alongside the existing test arguments,
ensuring the command uses the committed `apps/carrier/src-tauri/Cargo.lock`
dependency graph.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro Plus
Run ID: e8ae70e5-c76e-4e9c-ad1e-227299e832f3
📒 Files selected for processing (3)
.github/workflows/ts-suites.ymlapps/carrier/src-tauri/src/lib.rsscripts/ts-suites.mjs
| # Retry: this is the only apt-get in any workflow here, and a single transient mirror | ||
| # failure would block merge on a PR that merely touched packages/contracts/**. | ||
| for i in 1 2 3; do sudo apt-get update && break || sleep 5; done | ||
| sudo apt-get install --no-install-recommends \ |
There was a problem hiding this comment.
🩺 Stability & Availability | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
printf '--- workflow snippet ---\n'
if [ -f .github/workflows/ts-suites.yml ]; then
nl -ba .github/workflows/ts-suites.yml | sed -n '1,140p'
else
echo 'missing .github/workflows/ts-suites.yml'
fi
printf '\n--- bash behavior probe for retry pattern ---\n'
bash -c '
set -e
for i in 1 2 3; do
test "$i" != 3 || (echo "i=$i"; exit 77)
false && break || sleep 0
echo "loop_iteration_ended_i=$i"
done
echo "after_loop_status=$?"
' || echo "script_exited_status=$?"Repository: Harborline-Software/shipyard
Length of output: 257
🏁 Script executed:
#!/bin/bash
set -euo pipefail
printf '%s\n' '--- workflow snippet ---'
if [ -f .github/workflows/ts-suites.yml ]; then
nl -ba .github/workflows/ts-suites.yml | sed -n '1,160p'
else
echo 'missing .github/workflows/ts-suites.yml'
fi
printf '%s\n' ''
printf '%s\n' '--- bash behavior probe for retry pattern ---'
bash -c '
set -e
for i in 1 2 3; do
test "$i" != 3 || (echo "i=$i"; exit 77)
false && break || sleep 0
echo "loop_iteration_ended_i=$i"
done
echo "after_loop_status=$?"
' || echo "script_exited_status=$?"
printf '%s\n' ''
printf '%s\n' '--- behavioral pattern probe: failed apt with exit 65 ---'
tmp="$(mktemp)"
cat > "$tmp" <<'BASH'
set -u
failed=0
for i in 1 2 3; do
failed=$((failed + 1))
if false; then
break
fi
if [ "$i" -eq 3 ]; then
echo "apt-get update failed after 3 attempts" >&2
exit 65
fi
sleep 0
done
echo "completed_with_failed_attempts=$failed"
BASH
set -euo pipefail
bash "$tmp" || echo "script_exited_status=$?"
rm -f "$tmp"Repository: Harborline-Software/shipyard
Length of output: 231
🏁 Script executed:
#!/bin/bash
set -euo pipefail
printf '%s\n' '--- workflow snippet ---'
if [ -f .github/workflows/ts-suites.yml ]; then
awk '{printf "%6d %s\n", NR, $0}' .github/workflows/ts-suites.yml | sed -n '1,180p'
else
echo 'missing .github/workflows/ts-suites.yml'
fi
printf '%s\n' ''
printf '%s\n' '--- behavioral pattern probe: failed apt with exit 65 ---'
tmp="$(mktemp)"
cat > "$tmp" <<'BASH'
set -u
failed=0
for i in 1 2 3; do
failed=$((failed + 1))
if false; then
break
fi
if [ "$i" -eq 3 ]; then
echo "apt-get update failed after 3 attempts" >&2
exit 65
fi
sleep 0
done
echo "completed_with_failed_attempts=$failed"
BASH
set -euo pipefail
bash "$tmp" || echo "script_exited_status=$?"
rm -f "$tmp"
printf '%s\n' ''
printf '%s\n' '--- installed commands list ---'
for cmd in awk sed cat bash diff echo; do
command -v "$cmd" || echo "$cmd not found"
doneRepository: Harborline-Software/shipyard
Length of output: 8869
issue [blocking]: Fail closed when apt retries run out.
apt-get update && break || sleep 5 returns success after the third failed update because sleep is the final command. The workflow then continues with failed package indexes. Make the third failure exit non-zero.
Proposed retry fix
- for i in 1 2 3; do sudo apt-get update && break || sleep 5; done
+ for i in 1 2 3; do
+ if sudo apt-get update; then
+ break
+ fi
+ if [ "$i" -eq 3 ]; then
+ echo "apt-get update failed after 3 attempts" >&2
+ exit 1
+ fi
+ sleep 5
+ done📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| # Retry: this is the only apt-get in any workflow here, and a single transient mirror | |
| # failure would block merge on a PR that merely touched packages/contracts/**. | |
| for i in 1 2 3; do sudo apt-get update && break || sleep 5; done | |
| sudo apt-get install --no-install-recommends \ | |
| # Retry: this is the only apt-get in any workflow here, and a single transient mirror | |
| # failure would block merge on a PR that merely touched packages/contracts/**. | |
| for i in 1 2 3; do | |
| if sudo apt-get update; then | |
| break | |
| fi | |
| if [ "$i" -eq 3 ]; then | |
| echo "apt-get update failed after 3 attempts" >&2 | |
| exit 1 | |
| fi | |
| sleep 5 | |
| done | |
| sudo apt-get install --no-install-recommends \ |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In @.github/workflows/ts-suites.yml around lines 91 - 94, Update the apt-get
retry loop in the workflow so exhausting all three attempts exits non-zero
instead of succeeding because sleep is the final command. Preserve the
five-second delay between attempts, but explicitly fail after the final
unsuccessful apt-get update before continuing to package installation.
Source: Path instructions
| sudo apt-get install --no-install-recommends \ | ||
| libwebkit2gtk-4.1-dev \ | ||
| libxdo-dev \ | ||
| libssl-dev \ | ||
| libayatana-appindicator3-dev \ | ||
| librsvg2-dev |
There was a problem hiding this comment.
🩺 Stability & Availability | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "Workflow context:"
if [ -f .github/workflows/ts-suites.yml ]; then
nl -ba .github/workflows/ts-suites.yml | sed -n '1,180p'
else
echo ".github/workflows/ts-suites.yml not found"
fi
echo
echo "Search for apt-get install in workflows:"
rg -n "apt-get install|apt install|pnpm ts-suites|paths-filter" .github/workflows -S || true
echo
echo "Check whether sudo pip install has a non-interactive flag nearby (for related install context):"
rg -n "sudo apt-get install --no-install-recommends|sudo pip install|DEBIAN_FRONTEND|wget -nv --no-check-certificate" .github/workflows/ts-suites.yml -S || trueRepository: Harborline-Software/shipyard
Length of output: 224
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "Find workflow file:"
find .github/workflows -maxdepth 1 -name 'ts-suites.yml' -print 2>/dev/null || true
echo
echo "Workflow context using sed:"
if [ -f .github/workflows/ts-suites.yml ]; then
sed -n '1,180p' .github/workflows/ts-suites.yml
else
echo ".github/workflows/ts-suites.yml not found"
fi
echo
echo "Search for apt-get install in workflows:"
grep -RInE "apt-get install|apt install|pnpm ts-suites|paths-filter" .github/workflows || true
echo
echo "Check non-interactive flags nearby:"
grep -RInE "DEBIAN_FRONTEND|wget -nv --no-check-certificate|sudo apt-get install --no-install-recommends|sudo pip install" .github/workflows/ts-suites.yml .github/workflows 2>/dev/null || trueRepository: Harborline-Software/shipyard
Length of output: 9455
🌐 Web query:
Tauri v2 CI prerequisites debian ubuntu apt-get install -y libwebkit2gtk-4.1-dev libxdo-dev libssl-dev libayatana-appindicator3-dev librsvg2-dev
💡 Result:
To set up a development environment for Tauri v2 on Debian or Ubuntu, you must install the required system dependencies [1]. The official documentation recommends the following command [1]: sudo apt update sudo apt install libwebkit2gtk-4.1-dev \ build-essential \ curl \ wget \ file \ libxdo-dev \ libssl-dev \ libayatana-appindicator3-dev \ librsvg2-dev Key points regarding these prerequisites: 1. WebKitGTK 4.1: Tauri v2 requires libwebkit2gtk-4.1-dev [1][2]. This dependency is available in modern distributions such as Debian Bookworm (12+) and Ubuntu 22.04+ [3][4]. Attempting to build on older distributions that do not provide this package will result in build failures [5]. 2. libxdo-dev: This library is necessary for enabling features like Cut, Copy, Paste, and Select All in menus [6][7]. While older versions of documentation may have omitted it, it is currently included in the official prerequisite list [1][6]. 3. Build Environment: Because system library versions (specifically glibc) can impact compatibility, it is recommended to build your application on a distribution version that matches your target, or to use Docker/GitHub Actions to maintain a consistent environment [3][8]. Always ensure your system package information is updated via sudo apt update before attempting to install these packages [1][4].
Citations:
- 1: https://v2.tauri.app/start/prerequisites/
- 2: [bug] Tauri v2 has constrained compatibility on Linux tauri-apps/tauri#9039
- 3: https://v2.tauri.app/distribute/debian/
- 4: https://tauri.app/assets/learn/community/HTML_CSS_JavaScript_and_Rust_for_Beginners_A_Guide_to_Application_Development_with_Tauri.pdf
- 5: Hi, v2.0 alpha on linux compile problem tauri-apps/tauri#8492
- 6: [request] Add libxdo-dev to Linux prerequisites tauri-apps/tauri-docs#2230
- 7: https://docs.rs/tauri/latest/x86_64-linux-android/tauri/
- 8: https://v2.tauri.app/distribute/appimage/
🌐 Web query:
Tauri v2 tests webdriver CI apt-get install -y libwebkit2gtk-4.1-dev libxdo-dev libssl-dev libayatana-appindicator3-dev librsvg2-dev ubuntu-22.04
💡 Result:
For Tauri v2 WebDriver testing in a Linux CI environment (specifically Ubuntu 22.04), you must install specific system dependencies to support the Tauri application's runtime and the tauri-driver tool [1][2]. To set up your CI environment, use the following apt-get installation command [1]: sudo apt-get update && sudo apt-get install -y \ libwebkit2gtk-4.1-dev \ libayatana-appindicator3-dev \ webkit2gtk-driver \ xvfb Explanation of the key components: - libwebkit2gtk-4.1-dev: Required as the underlying WebView engine for Tauri v2 [3][4][5]. - libayatana-appindicator3-dev: Often necessary for system tray and notification support in Tauri apps [6][3]. - webkit2gtk-driver: The native WebDriver server for Linux (WebKitGTK) that tauri-driver uses to interact with your application [1][2][7]. - xvfb: A virtual framebuffer X server commonly required in CI environments to run GUI applications (or WebViews) in "headless" mode without a physical display [1]. While your query mentioned libxdo-dev, libssl-dev, and librsvg2-dev (which are standard build-time dependencies for Tauri applications [6]), they are not explicitly listed in the minimal CI pipeline for WebDriver testing provided by official Tauri v2 documentation [1]. However, if your application requires these for building, you may include them in your installation step [6]. Ensure that the tauri-driver binary is installed in your CI environment as well, typically via cargo install tauri-driver --locked [2][7].
Citations:
- 1: https://v2.tauri.app/develop/tests/webdriver/ci/
- 2: https://v2.tauri.app/develop/tests/webdriver/manual-setup/
- 3: https://v2.tauri.app/distribute/debian/
- 4: https://v2.tauri.app/blog/tauri-2-0-0-alpha-3/
- 5: https://docs.crabnebula.dev/cloud/ci/tauri-v2-workflow/
- 6: https://v2.tauri.app/start/prerequisites/
- 7: https://docs.rs/crate/tauri-driver/latest
🌐 Web query:
apt-get install without -y will prompt confirmation prompt
💡 Result:
To run apt-get install or other commands without being stopped by confirmation prompts, you can use the following methods depending on the type of prompt you are encountering. Basic Confirmation Prompts For the standard "Do you want to continue [Y/n]?" prompt, use the --yes (or -y) flag [1][2]: sudo apt-get install -y <package_name> Handling Configuration and Interactive Prompts Some packages require additional input during installation (e.g., configuring services). To automate these and force the system to use default answers, set the DEBIAN_FRONTEND environment variable to noninteractive [3][4]: sudo DEBIAN_FRONTEND=noninteractive apt-get install -y <package_name> Advanced Configuration Management If an installation stops due to configuration file conflicts, you can instruct the underlying dpkg utility to keep existing files or use default options using the --option (or -o) flag [4][5]: sudo DEBIAN_FRONTEND=noninteractive apt-get install -y -o Dpkg::Options::='--force-confdef' -o Dpkg::Options::='--force-confold' <package_name> Additional Automation Techniques - Pre-seeding: For more complex packages that require specific answers, you can use debconf-set-selections to pre-seed the desired responses into the debconf database before running the installation [6]. - Ensure apt-utils is installed: If non-interactive mode is still prompting, ensure the apt-utils package is installed on the system, as it is often required for proper non-interactive operation [7]. Top results: [3][1][7][4][6][5]
Citations:
- 1: https://superuser.com/questions/164553/automatically-answer-yes-when-using-apt-get-install
- 2: https://www.debian.org/doc/manuals/apt-guide/ch2.en.html
- 3: https://askubuntu.com/questions/556385/how-can-i-install-apt-packages-non-interactively
- 4: https://sleeplessbeastie.eu/2023/10/30/how-to-install-packages-non-interactively-using-apt/
- 5: https://stackoverflow.com/questions/24926985/force-no-prompt-in-apt-get-upgrade-or-apt-get-dist-upgrade
- 6: https://stackoverflow.com/questions/702248/how-can-i-automate-dpkg-apt-get
- 7: https://unix.stackexchange.com/questions/786421/how-to-answer-questions-from-apt-install-without-getting-prompted
issue [blocking]: Make the Tauri prerequisite installation non-interactive.
apt-get install can prompt for sudo apt-get install --no-install-recommends, which stops the required CI check before pnpm ts-suites runs. Add the non-interactive flag to the install command.
Proposed install change
- sudo apt-get install --no-install-recommends \
+ sudo DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends \📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| sudo apt-get install --no-install-recommends \ | |
| libwebkit2gtk-4.1-dev \ | |
| libxdo-dev \ | |
| libssl-dev \ | |
| libayatana-appindicator3-dev \ | |
| librsvg2-dev | |
| sudo DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends \ | |
| libwebkit2gtk-4.1-dev \ | |
| libxdo-dev \ | |
| libssl-dev \ | |
| libayatana-appindicator3-dev \ | |
| librsvg2-dev |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In @.github/workflows/ts-suites.yml around lines 94 - 99, Update the
prerequisite installation command in the workflow step to include apt-get’s
non-interactive option while preserving the existing package list and
no-recommends behavior, so the Tauri dependencies install without prompting
before pnpm ts-suites runs.
| run( | ||
| 'test Carrier Tauri Rust host', | ||
| 'cargo', | ||
| ['test', '--manifest-path', 'apps/carrier/src-tauri/Cargo.toml'], | ||
| '.', | ||
| { TAURI_CONFIG: '{"bundle":{"externalBin":[]}}' }, |
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "== files =="
git ls-files | rg '^scripts/ts-suites\.mjs$|^apps/carrier/src-tauri/(Cargo\.toml|Cargo\.lock)$|^_shared/engineering/code-review-policy\.yaml$|\.github/workflows|package\.json|pnpm-lock' || true
echo
echo "== ts-suites around command =="
if [ -f scripts/ts-suites.mjs ]; then
nl -ba scripts/ts-suites.mjs | sed -n '110,175p'
fi
echo
echo "== references to test Carrier / Cargo.lock =="
rg -n "test Carrier|Cargo.lock|cargo test|cargo update|src-tauri/Cargo.toml|cargo --locked|--locked" .github scripts apps/carrier/src-tauri -S || true
echo
echo "== Cargo manifest deps =="
if [ -f apps/carrier/src-tauri/Cargo.toml ]; then
nl -ba apps/carrier/src-tauri/Cargo.toml | sed -n '1,140p'
fi
echo
echo "== Cache/dependency cache references =="
rg -n "cache.*Cargo|cargo-cache|target/.cargo|Cargo.lock|actions/cache" .github scripts -S || true
echo
echo "== package scripts =="
for f in package.json pnpm-workspace.yaml; do
[ -f "$f" ] && nl -ba "$f" | sed -n '1,160p'
doneRepository: Harborline-Software/shipyard
Length of output: 2383
🌐 Web query:
Cargo test --locked documentation missing Cargo.lock will fail exit code
💡 Result:
For most Cargo commands that utilize the --locked flag—including cargo test—the presence of a Cargo.lock file is strictly required [1][2][3]. If the lock file is missing, Cargo will exit with an error [1][2]. The official Cargo documentation explicitly states that when --locked is used, Cargo asserts that the exact same dependencies and versions are used as when the existing Cargo.lock file was originally generated [1][4]. Consequently, Cargo will exit with an error if the lock file is missing or if the current dependency resolution requires updating the lock file [1][2][4]. While this behavior is consistent across most commands (such as cargo test, cargo build, and cargo update), there have been historical discrepancies and discussions regarding specific edge cases, particularly with cargo install, where behaviors or documentation may occasionally appear inconsistent [5][6]. However, for standard project operations like cargo test, the requirement is definitive: a missing Cargo.lock file results in a failure [1][2].
Citations:
- 1: https://doc.rust-lang.org/cargo/commands/cargo-test.html
- 2: https://doc.rust-lang.org/1.67.1/cargo/commands/cargo-test.html
- 3: https://dev-doc.rust-lang.org/beta/cargo/commands/cargo-update.html
- 4: https://doc.rust-lang.org/stable/cargo/commands/cargo-update.html
- 5: Contradictory documentation for
cargo install --lockedbehavior with missing lock file rust-lang/cargo#16649 - 6:
cargo install --lockedis not really locked rust-lang/cargo#9289
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "== ts-suites.mjs size and command section =="
wc -l scripts/ts-suites.mjs
sed -n '120,155p' scripts/ts-suites.mjs
echo
echo "== Cargo files existing/tracked =="
for f in scripts/ts-suites.mjs apps/carrier/src-tauri/Cargo.lock apps/carrier/src-tauri/Cargo.toml; do
if [ -f "$f" ]; then
echo "-- $f --"
sed -n '1,180p' "$f"
else
echo "-- missing: $f --"
fi
done
echo
echo "== cargo test references =="
rg -n "cargo test|Cargo.lock|--locked|src-tauri/Cargo.toml|cache.*Cargo|actions/cache" \
.github/workflows scripts apps/carrier . \
-g '!**/pnpm-lock.yaml' -g '!**/Cargo.lock' || true
echo
echo "== workflow names referencing ts suites or carrier cargo =="
rg -n "ts-suites|test Carrier|Carrier Tauri|cargo" .github/workflows scripts/ts-suites.mjs -S || trueRepository: Harborline-Software/shipyard
Length of output: 50384
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "== ts-suites workflow relevant action setup =="
sed -n '1,145p' .github/workflows/ts-suites.yml
echo
echo "== read-only Cargo lockfile/hashfiles facts =="
python3 - <<'PY'
from pathlib import Path
import re, hashlib, subprocess
root = Path('.')
for path in ["scripts/ts-suites.mjs", ".github/workflows/ts-suites.yml", "apps/carrier/src-tauri/Cargo.toml", "apps/carrier/src-tauri/Cargo.lock", "packages/contracts/rust/Cargo.lock"]:
p = root / path
print(f"{p}: exists={p.exists()}")
mjs = root / "scripts/ts-suites.mjs"
print("cargo args in scripts/ts-suites.mjs:")
payload = mjs.read_text()
for line in payload.splitlines():
if "apps/carrier/src-tauri/Cargo.toml" in line:
print(" ", line.strip())
yml = (root / ".github/workflows/ts-suites.yml").read_text()
action = yml.split("runTsSuites:")[-1].split(" ts-suite")[0]
print("ts-suites action key/hashfiles:")
for line in action.splitlines()[40:100]:
if "hashFiles" in line or "restore-keys" in line or "cargo" in line:
print(line)
cargo = (root / "apps/carrier/src-tauri/Cargo.toml").read_text()
direct_names = set()
for m in re.finditer(r'^([a-z0-9_-]+)\s*=\s*\{', cargo, re.M):
direct_names.add(m.group(1))
lock = (root / "apps/carrier/src-tauri/Cargo.lock").read_text()
for entry in re.finditer(r'\[\[package]]\n', lock):
start = entry.start()
block = lock[start:lock.find("\n[[package]]", start + 1) or len(lock)]
name = re.search(r'^name\s*=\s*"([^"]+)"', block, re.M)
if name:
version = re.search(r'^[ ]*version\s*=\s*"([^"]+)"', block, re.M)
direct = name.group(1) in direct_names
if version and direct and name.group(1) in ("tauri", "tauri-build", "tauri-plugin-shell", "tauri-plugin-http", "serde_json", "reqwest", "keyring"):
print("direct workspace dependency:", name.group(1), "version", version.group(1), "direct", direct)
PYRepository: Harborline-Software/shipyard
Length of output: 7896
issue [blocking]: Pin the Carrier Tauri Cargo test to the committed lockfile.
This required Rust test omits --locked, so Cargo can regenerate dependency selections while apps/carrier/src-tauri/Cargo.lock exists. Add --locked so CI must use the committed dependency graph rather than an updated one.
Proposed command change
- ['test', '--manifest-path', 'apps/carrier/src-tauri/Cargo.toml'],
+ ['test', '--locked', '--manifest-path', 'apps/carrier/src-tauri/Cargo.toml'],📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| run( | |
| 'test Carrier Tauri Rust host', | |
| 'cargo', | |
| ['test', '--manifest-path', 'apps/carrier/src-tauri/Cargo.toml'], | |
| '.', | |
| { TAURI_CONFIG: '{"bundle":{"externalBin":[]}}' }, | |
| run( | |
| 'test Carrier Tauri Rust host', | |
| 'cargo', | |
| ['test', '--locked', '--manifest-path', 'apps/carrier/src-tauri/Cargo.toml'], | |
| '.', | |
| { TAURI_CONFIG: '{"bundle":{"externalBin":[]}}' }, |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@scripts/ts-suites.mjs` around lines 141 - 146, Update the `run` invocation
for `test Carrier Tauri Rust host` to include Cargo’s `--locked` flag alongside
the existing test arguments, ensuring the command uses the committed
`apps/carrier/src-tauri/Cargo.lock` dependency graph.
Source: Path instructions
Kept HEAD's new step-5 addition (compile + cargo test the Carrier Tauri Rust host, with TAURI_CONFIG externalBin suppressed so the run does not need the platform sidecar) plus its explanatory comment, positioned right after the contracts Rust projection test as authored. Took origin/main's package rename of the four test() labels and filter targets from shipyard/* to harborline-software/* (an unrelated rename that landed on main after this branch forked). The workflow file (.github/workflows/ts-suites.yml) auto-merged cleanly: HEAD's raised timeout, apt-get prerequisites and cargo cache steps combined with origin/main's added ui-react/ui-adapters-react path-filter entries. For verification, a fresh pnpm install --frozen-lockfile showed no lockfile drift, node --check on the merged script found no syntax error, and a label-by-label diff confirmed every run() step present on origin/main is still present post-merge with exactly one addition. The new step's own command was then run directly — cargo test --manifest-path apps/carrier/src-tauri/Cargo.toml with TAURI_CONFIG set to disable externalBin — which compiles the full Tauri host and passes 45 of 47 tests (2 intentionally ignored, desktop-only), including the protocol contract test that consumes the complete generated command inventory. The unaffected contracts Rust projection suite also still passes (2/2), as a sanity check on the earlier step. Refs: shipyard#3541
1 participant
ci(carrier): compile and test the Tauri host, which no job touched
Nothing in CI ran cargo for
apps/carrier/src-tauri. Two consequences: the host's own protocolcontract tests never executed, and the generated-contract path dependency added by the
language-agnostic contracts work was never compiled, so a break in the Rust host's consumption of
that projection could not be caught.
The inert workflow whose path globs point at a directory that does not exist in this repository is
an intentional migration stub, and is left alone — it is documented as such rather than being
accidental coverage.
The host's cargo test now runs in the existing carrier/hull job, whose path filter already covers
the relevant trees.
TAURI_CONFIGsuppresses only the bundle-time external sidecar; this compilesand tests the host without packaging or launching the desktop app.
A cargo cache is included. Without it every run recompiles the webkit bindings from scratch, and a
slow required check is how people learn to ignore CI.
The stale comment claiming CI does not test this crate is replaced with the actual coverage
boundary.
Proven by mutation: pointing the host at a generated constant that does not exist fails compilation
in the new step, so the step really does compile the host's consumption of the projection.
Honest limitation: GitHub Actions itself was not exercised. The commands were verified locally on
macOS; the Ubuntu runner path — the apt prerequisites in particular — is unverified until this runs.
Watch the first execution rather than assuming it.
Co-Authored-By: Claude Opus 5 (1M context) noreply@anthropic.com
Summary by CodeRabbit
Tests
Chores