Verifier mutation scores

For each sample task, one plausible bug at a time was planted in the reference solution (operator and comparison flips, constant tweaks, dropped statements, negated conditions, ...) and graded against the task's cases. A killed mutant fails at least one case; a survivor passes them all (an equivalent program, or a gap in the cases). Mutants that do not compile or do not change the program are invalid. Generated by tools/mutation_table.py.

Task	Mutants	Killed	Survived	Invalid	Score
py-balanced-brackets	24	24	0	0	100.0%
py-flatten	14	14	0	0	100.0%
py-gcd-lcm	11	11	0	0	100.0%
py-matrix-rotate	12	12	0	0	100.0%
py-merge-intervals	14	14	0	0	100.0%
py-parse-duration	21	18	1	2	94.7%
py-rle-encode	17	17	0	0	100.0%
py-roman	40	40	0	0	100.0%
py-semver-compare	35	30	2	3	93.8%
py-slugify	11	11	0	0	100.0%
py-two-sum	6	4	0	2	100.0%
py-valid-ipv4	30	30	0	0	100.0%
py-word-frequency	4	4	0	0	100.0%
All tasks	239	229	3	7	98.7%

Survivors (3)

Each survivor is either behaviourally equivalent to the original or points at a case the task does not have. They are listed so that a human can decide which.

py-parse-duration, constant bool (line 13)

@@ -11 +11 @@
-    return sum((int(g) * s for g, s in zip(match.groups(), _SECONDS, strict=True) if g is not None))
+    return sum((int(g) * s for g, s in zip(match.groups(), _SECONDS, strict=False) if g is not None))

py-semver-compare, comparison flip (line 26)

@@ -23 +23 @@
-    return -1 if pre_a < pre_b else 1
+    return -1 if pre_a <= pre_b else 1

py-semver-compare, comparison flip (line 19)

@@ -16 +16 @@
-        return -1 if nums_a < nums_b else 1
+        return -1 if nums_a <= nums_b else 1