A pull enters the merge queue only after the full suite passes on its current head merged with main (not the impact-mapped subset) #7616

Measured 2026-10-27 23:40Z–01:13Z:

• 15 merge-queue gates ran, and 11 were red.
• 5 of the reds were one test broken by one pull (#7566), which passed its own check.
• Gate time: p50 562 s, max 849 s.
• Landed: about 20 an hour, with 45 approved pulls waiting.

The owner: "i need this to be faster. much faster. and stabily"

Cause: a pull's check runs the impact-mapped subset, and the gate runs everything, so breakage is found in the queue, where a red costs a batch of up to 8.

Build:

• Before admission, run the full suite (the same command and environment as the gate) on the pull's head merged with current main. Admit it only when that's green.
• Re-run it when main has moved by more than N landings since.
• Use the same sharding as the gate (the sibling issue).
• A red here goes back to the builder with the failing test named, at no cost to anyone else.

Tests:

• A pull that breaks an unrelated test is never admitted.
• An admitted pull whose proof is stale (main moved) is re-proved before its batch.

Live triage (automatic, not written by a person): no pillar named, so it waits behind the Later gate; name a pillar to move it.