Skip to content

test(codex): unskip full-turn coverage #1691

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
NathanFlurry wants to merge 1 commit into from
+22 −8
Closed

test(codex): unskip full-turn coverage #1691

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 16 additions & 2 deletions docs-internal/registry-parity-worklist.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -546,8 +546,22 @@ real e2e tests that prove Linux-parity behavior — not smoke tests.
`2026-07-08T12-46-11-0700-item11-core-tsc-duckdb-final-file-after-refresh.log`;
`2026-07-08T12-46-10-0700-item11-duckdb-biome-check-final-file-after-refresh.log`.
Rev: `qrwnvouk`.
- **codex — TODO.** Needs the remaining real full-turn coverage un-skipped after
#9.
- **codex — DONE.** Un-skipped the remaining real `codex-exec --session-turn`
full-turn coverage. The suite now proves the model turn, on-request shell
tool call, real subprocess filesystem side effect, and adapter-supplied
history replay in one VM-backed file with 4/4 passing. For this coverage
rev, the package was staged from existing real 29 MB `codex`/`codex-exec`
artifacts because rebuilding the external Codex fork in this checkout hit
dependency gating failures (`path-dedot`, then `tokio`/`rustls-native-certs`);
the failed rebuild logs are kept as proof. Proof:
`2026-07-08T12-52-54-0700-item11-codex-cli-package-build-from-prior-artifacts-after-install.log`;
`2026-07-08T12-54-08-0700-item11-codex-fullturn-final-unskipped.log`;
`2026-07-08T12-54-42-0700-item11-core-tsc-codex-final.log`;
`2026-07-08T12-54-42-0700-item11-codex-biome-check-final.log`;
rebuild blockers:
`2026-07-08T12-48-46-0700-item11-codex-required-build-fresh-cargo-home.log`,
`2026-07-08T12-50-38-0700-item11-codex-required-build-path-dedot-cfg.log`.
Rev: `ryqtvoqv`.
- **rev:** one per command, e.g. `test(duckdb): real analytical-SQL e2e; un-skip`

### 12. No tests at all — 9 software + 5 agents
Expand Down
12 changes: 6 additions & 6 deletions packages/core/tests/codex-fullturn.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -153,9 +153,7 @@ describe("codex full turn (real codex agent in the VM, mock OpenAI Responses)",
expect(stdout).toContain('"type":"done"');
}, 70000);

// Skipped until the WASI Codex tool watcher reliably completes file-writing
// subprocesses; the simple real subprocess tool path above remains active.
test.skip("shell tool runs a REAL subprocess with an observable filesystem side effect", async () => {
test("shell tool runs a REAL subprocess with an observable filesystem side effect", async () => {
// Proves codex's exec tool spawns a real subprocess via the secure-exec
// host_process bridge (not a mocked/gated stub): the model asks to run a
// shell command that WRITES A FILE, we approve it, and after the turn we
Expand Down Expand Up @@ -215,6 +213,10 @@ describe("codex full turn (real codex agent in the VM, mock OpenAI Responses)",
JSON.stringify({ decision: "allow" }) +
"\n";
await vm.execArgv("mkdir", ["-p", "/root/.codex"]);
await vm.writeFile(
"/root/.codex/config.toml",
new TextEncoder().encode(codexConfig),
);
await vm.writeFile(sourcePath, new TextEncoder().encode(marker));
const r = await vm.execArgv("codex-exec", ["--session-turn"], {
timeout: 45000,
Expand All @@ -236,9 +238,7 @@ describe("codex full turn (real codex agent in the VM, mock OpenAI Responses)",
}
}, 70000);

// Skipped until the WASI session-turn wrapper reliably completes resumed-history
// turns instead of hanging after the mock response.
test.skip("replays adapter-supplied history on a resumed multi-turn session", async () => {
test("replays adapter-supplied history on a resumed multi-turn session", async () => {
const { stdout, requests } = await runSessionTurn(
[finalText("the answer is 4")],
{
Expand Down