fix(release): publish sharpebench-edge before sharpebench-wasm#9
Conversation
v0.0.10 wired sharpebench-edge into sharpebench-wasm, but the publish loop still listed wasm before edge, so wasm@0.0.10 failed to resolve edge@0.0.10 (not yet published). Move edge ahead of wasm. The loop is idempotent, so re-running completes the 4 crates that the first v0.0.10 run did not reach (edge, wasm, harness, sharpebench); the 6 already at 0.0.10 are skipped.
📝 WalkthroughWalkthroughIn the release workflow's crate publishing loop, ChangesRelease Workflow Crate Publish Order
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~2 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches🧪 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 |
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 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/release.yml:
- Around line 91-95: The release fallback publish flow is out of sync with the
documented package order because scripts/publish.sh still omits
sharpebench-stats and sharpebench-edge, which can break sharpebench-wasm
publishing. Update scripts/publish.sh to use the same package order as the
release workflow (including sharpebench-stats and sharpebench-edge before
sharpebench-wasm) or remove scripts/publish.sh as a supported release path if it
should no longer be used.
🪄 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: defaults
Review profile: CHILL
Plan: Pro
Run ID: aca81a72-26c2-492c-9dca-1ac4b63ebf90
📒 Files selected for processing (1)
.github/workflows/release.yml
| # harness on sim; sharpebench on all). edge depends on stats and is itself a dep of | ||
| # wasm + the CLI, so it MUST come before wasm. publish=false members are omitted. | ||
| for c in sharpebench-stats sharpebench-edge sharpebench-core sharpebench-protocol \ | ||
| sharpebench-attest sharpebench-sim sharpebench-leaderboard sharpebench-wasm \ | ||
| sharpebench-harness sharpebench; do |
There was a problem hiding this comment.
🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "== release workflow crate order =="
sed -n '84,105p' .github/workflows/release.yml
echo
echo "== publish script crate order =="
sed -n '1,80p' scripts/publish.sh
echo
echo "== references to scripts/publish.sh =="
rg -n 'scripts/publish\.sh|publish\.sh' . github .github docs README.mdRepository: general-liquidity/sharpebench
Length of output: 3847
Keep scripts/publish.sh aligned with the release order scripts/publish.sh still skips sharpebench-stats and sharpebench-edge, so the documented fallback publish flow can break on sharpebench-wasm. Update the script in this PR or remove it as a supported path.
🤖 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/release.yml around lines 91 - 95, The release fallback
publish flow is out of sync with the documented package order because
scripts/publish.sh still omits sharpebench-stats and sharpebench-edge, which can
break sharpebench-wasm publishing. Update scripts/publish.sh to use the same
package order as the release workflow (including sharpebench-stats and
sharpebench-edge before sharpebench-wasm) or remove scripts/publish.sh as a
supported release path if it should no longer be used.
v0.0.10 made
sharpebench-wasmdepend onsharpebench-edge, but the publish loop listedwasmbeforeedge, sowasm@0.0.10couldn't resolveedge@0.0.10. Reorder edge ahead of wasm. Idempotent loop completes the 4 crates the first run missed (edge/wasm/harness/sharpebench); the 6 already at 0.0.10 are skipped on re-run.Summary by CodeRabbit
sharpebench-edgeearlier in the sequence.