fix(devnet): re-deploy contracts when artifacts outpace running chain

[zsculac]

Summary

Two fixes to scripts/devnet.sh so a long-running Hardhat node doesn't silently serve stale bytecode after a contract recompile.

1. Stale-bytecode guard in start_hardhat

When Hardhat is left running across edits to .sol sources, on-chain bytecode lags the recompiled artifacts on disk. Next devnet start short-circuits on the alive PID, then deploy_contracts skips on the still-present .devnet/hardhat/deployed marker. Daemons connect to addresses whose code predates the latest source; view methods added since that deploy revert with "function selector was not recognized." Ethers surfaces this as require(false) with no return data, masking the real cause.

On a 6-node devnet this manifested as FATAL: only 0/4 core node(s) staked because the staking loop's CSS.getNodeStakeV10 probe failed for every node.

The fix compares artifact JSON mtimes to the marker. If any artifact is newer, kill Hardhat so the existing fresh-start path (clears marker + localhost_contracts.json + hardhat_contracts.json) fires and deploy_contracts redeploys onto a fresh chain. No-op when artifacts are unchanged — keeps the fast path intact for the common edit-nothing-restart-devnet flow.

2. Escape backticks in node -e JS comments

Five JS comments inside the staking heredoc (lines 742, 868, 881, 898, 899) had unescaped backticks (`continue`, `nonce++`, `EVMChainAdapter.ensureProfile()`, `getNodeStakeV10(idId)`, `balanceOf(opSigner.address)`). Bash command-substituted them inside the surrounding node -e "..." double-quoted string, so every devnet start printed:

scripts/devnet.sh: line 694: continue: only meaningful in a `for', `while', or `until' loop
scripts/devnet.sh: line 694: nonce++: command not found
scripts/devnet.sh: line 694: syntax error near unexpected token `idId'

Text was inside // comments so the staking script still ran, but the diagnostic noise drowned out real failures (e.g. the stale-bytecode revert above). Lines 730-733 and 768 already use ``` correctly; these five were oversights.

Test plan

• bash -n scripts/devnet.sh syntax check
• Fresh devnet stop && devnet start 6 no longer prints bash backtick errors
• Confirmed contract redeploy on artifacts-newer scenario by leaving long-running hardhat alive across recompile, then devnet start — stale-bytecode guard fires, contracts redeployed fresh
• Reviewer: walk through fast path (artifacts unchanged) — start_hardhat short-circuits with no marker mtime check overhead

Out of scope

A pre-existing nonce race between the daemon's ensureProfile() auto-stake and the bootstrap script's stake loop (both target the same op wallet within seconds of each other) was masked by the bash syntax errors. With these fixes the race surfaces as Stake failed for node N: nonce has already been used. Filing separately — not introduced by this PR.

🤖 Generated with Claude Code

[github-actions]

🟡 Issue: This restart path removes hardhat.pid and immediately starts a replacement even if the old Hardhat process is still alive after the 5s wait. On a slow shutdown, the new hardhat node can fail to bind :$HARDHAT_PORT, turning the stale-artifact recovery into a flaky devnet start. Re-check kill -0 after the loop and either keep waiting/escalate the kill or return non-zero before deleting the pidfile and launching a new node.