Why it matters

When thousands of endpoints fail in several scenarios, engineers need to know which scenario to fix against. If one scenario is worse everywhere, fixing it fixes the others. If not, each scenario owns some endpoints and the work has to be split. Dominance and ownership make that visible.

Worked example

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

slack = {
    "S1_ss_cworst":   [0.05, 0.12, -0.02, 0.30, 0.08],
    "S2_ss_rcworst":  [0.07, 0.10,  0.01, 0.28, 0.20],
    "S3_ss_cold_cw":  [0.04, 0.15,  0.03, 0.35, 0.09],
    "S4_tt":          [0.20, 0.25,  0.18, 0.40, 0.30],
}
names = list(slack)
def dominates(a, b):
    return all(x <= y for x, y in zip(slack[a], slack[b]))
dominated = sorted({b for a in names for b in names if a != b and dominates(a, b)})
assert dominated == ["S4_tt"]
assert not dominates("S1_ss_cworst", "S2_ss_rcworst") and not dominates("S2_ss_rcworst", "S1_ss_cworst")
owner = [min(names, key=lambda s: slack[s][i]) for i in range(5)]
assert owner == ["S3_ss_cold_cw", "S2_ss_rcworst", "S1_ss_cworst", "S2_ss_rcworst", "S1_ss_cworst"]
failing = [i for i in range(5) if slack[owner[i]][i] < 0]
assert failing == [2] and owner[2] == "S1_ss_cworst"
print("PASS: dominated", dominated, "; owners", owner, "; failing endpoint owned by", owner[2])

S4 (typical) is dominated by every slow scenario, as expected. No slow scenario dominates another: S1 is worse at endpoints 2 and 4, S2 at endpoints 1 and 3, S3 at endpoint 0. The only failing endpoint (index 2, slack -0.02 ns) is owned by S1, so the fix is judged in S1 and rechecked in the others. Endpoints owned by different scenarios need different fixes: a fix that helps in the cold corner may not help in the hot one.

Ownership table

What the review looks for

• A table of how many failing endpoints each scenario owns, so effort goes where it matters.

• Failing endpoints with different owners are triaged separately.

• After a fix, the owner and the runner-up scenario are both rechecked: the runner-up can become the new owner.

Related topics

• Pruning Scenarios