Repository navigation
docs(env): say that the URL variables do not move a server - #506
Conversation
The row for ECHIDNA_JULIA_COPROC_URL said to set one of the two URL variables when the coprocessor router and the ML sidecar both run. That is not enough: each variable sets only the address that the Rust client calls. The row now says to start one server on another port and then set its variable to match. Raised in review on the first pull request. Checked against src/julia/run_server.jl (port is the second argument) and src/julia/coprocessor.jl (start_coprocessor_server takes the port). Co-Authored-By: Claude <[email protected]> Claude-Session: https://claude.ai/code/session_0136P7Zmp1i8PGVHkq3u2iNz
Codex Review Summary: ✅ CompletedThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
|
Status of the red checks on head
I did not change a dependency here. A dependency bump in this workspace is a separate change with its own review, and a security finding stays open for the owner to decide. The merge of this PR is blocked until those two checks pass or the owner rules on them. Generated by Claude Code |
## Summary The daily `Cargo Audit` on `main` has failed since the 14 Nov 06:06 UTC scan, and with it every PR's `rust-ci / Cargo audit (security)` and `Dependency audit` jobs (for example #506, a docs-only change). The cause is a new advisory, not any PR: `RUSTSEC-2026-0538` (`GHSA-6pp4-4cxf-xf88`, "Receive loop can spin after a truncated datagram") affects `quinn-udp 0.5.14`, which is pinned in both `Cargo.lock` and `fuzz/Cargo.lock`. It is fixed in `>=0.5.16, <0.6.0`. ## Changes - `Cargo.lock`, `fuzz/Cargo.lock`: `cargo update -p quinn-udp --precise 0.5.16`. No other crate changes version. `windows-sys 0.60` and its `windows_*` target crates drop out, because `quinn-udp 0.5.16` no longer depends on them. Not changed: `RUSTSEC-2026-0486` (`rustix 1.1.4`, unsound) is still reported, but only as a warning, so it does not fail the gate. ## RSR Quality Checklist ### Required - [ ] Tests pass: left to CI (see Testing) - [x] Code is formatted: no source change - [x] Linter is clean: no source change - [x] No banned language patterns - [x] No `unsafe` blocks without `// SAFETY:` comments - [x] No banned functions - [x] SPDX license headers present on all new/modified source files (no source files changed) - [x] No secrets, credentials, or `.env` files included ## Testing - `bash scripts/ci/audit-dependencies.sh`, with `cargo-audit 0.22.2` installed by `scripts/ci/install-cargo-audit.sh` (checksum OK): `Cargo.lock: PASSED`, `fuzz/Cargo.lock: PASSED`, exit 0. Before the change, both said `FAILED (vulnerabilities found)` with `RUSTSEC-2026-0538`. - `cargo check --locked --workspace --all-targets` could not complete in my cloud container. The only errors are two build scripts that need system tools the container lacks: `echidna-wire` (`capnpc build failed: Failed to execute capnp --version: No such file or directory`) and `echidna-grpc` (its custom build command). Neither crate is touched by this change, and no Rust compile error was reported. The `rust-ci` build jobs on this PR are the real build test. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_0136P7ZoZwvZCfM387TZQ5kY Co-authored-by: Claude <[email protected]>
|
No actionable comments were generated in the recent review. Summary
Merge Risk:
|
|



Summary
Follow-up to echidna#505. The review on #505 found one imprecise sentence in the new row, and #505 merged before it was corrected.
The row said to set one of the two URL variables when the coprocessor router and the ML sidecar both run. That is not enough. Each variable sets only the address that the Rust client calls. It does not change the port that a server binds.
Changes
docs/ENV-VARS.adoc: the row forECHIDNA_JULIA_COPROC_URLnow says to start one server on another port (start_coprocessor_server(port), or the second argument ofsrc/julia/run_server.jl), then set its variable to match.Testing
Documentation only. Checked against
src/julia/run_server.jl(the port is the second argument) andsrc/julia/coprocessor.jl(start_coprocessor_servertakes the port).asciidoctor --failure-level=WARNon the file: clean.Generated by Claude Code