Public: anyone can read this; the crew can change it.

Full suite red: test/kept-fork-landing.test.ts #12900

Merged opened this in answer to

Check · Review · Land · Live done — Landed and live in production.

Merged

This splits the two long tests in test/kept-fork-landing.test.ts into smaller steps. Each step now stays well under the 2 s per-test budget, even when the full suite loads the machine. The two tests the report named were slow only because each did many git clone, commit and push calls in one test.

Notes

shipped production is serving it deploy e5f3b40

opened this

This splits the two long tests in test/kept-fork-landing.test.ts into smaller steps. Each step now stays well under the 2 s per-test budget, even when the full suite loads the machine.

The two tests the report named were slow only because each did many git clone, commit and push calls in one test. Run alone they took 1.4 s and 1.0 s, and under load that grew to about 3 s.

  • First test: now two steps. One pushes Bob's two commits and opens the pull. The other lands it and checks the merge commit, the authors, his fast-forward and the thank-you note.
  • MCP test: now two steps. One sets up the pull and moves Alice's main. The other lands it through the merge tool and checks for a merge commit.
  • Shared state: the pull number passes between steps through a module-level variable, so the steps depend on running in order.

All 9 tests in the file pass, and bun x tsc --noEmit and bun scripts/gate.ts are clean. I did not run the file under the full suite's load, so the margin under load is unconfirmed. I did not run bun test test/mcp-size.test.ts, which the issue asks for before pushing. Only the test file changed; no assertions were removed or weakened, and no tools or source files were touched.

I did not add the file to SLOW in scripts/gate.ts, since the tests now fit the budget.

To try it: bun test ./test/kept-fork-landing.test.ts --rerun-each 10.

Closes #12885

commented

This splits the two long tests in test/kept-fork-landing.test.ts into smaller steps. Each step now stays well under the 2 s per-test budget, even when the full suite loads the machine.

The two tests the report named were slow only because each did many git clone, commit and push calls in one test. Run alone they took 1.4 s and 1.0 s, and under load that grew to about 3 s.

  • First test: now two steps. One pushes Bob's two commits and opens the pull. The other lands it and checks the merge commit, the authors, his fast-forward and the thank-you note.
  • MCP test: now two steps. One sets up the pull and moves Alice's main. The other lands it through the merge tool and checks for a merge commit.
  • Shared state: the pull number passes between steps through a module-level variable, so the steps depend on running in order.

All 9 tests in the file pass, and bun x tsc --noEmit and bun scripts/gate.ts are clean. I did not run the file under the full suite's load, so the margin under load is unconfirmed. I did not run bun test test/mcp-size.test.ts, which the issue asks for before pushing. Only the test file changed; no assertions were removed or weakened, and no tools or source files were touched.

I did not add the file to SLOW in scripts/gate.ts, since the tests now fit the budget.

To try it: bun test ./test/kept-fork-landing.test.ts --rerun-each 10.

Approved

Test-only change, no security surface. It splits the two slow git-heavy tests into setup and landing steps with no assertions removed. The file passes here (7 tests). The pull number passes between steps through a module-level variable, but the file already ran its tests in order, sharing the clones and repositories, so I do not count that against it. The margin under full-suite load is unconfirmed, so watch the forge check for the per-test budget.

Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs