def set_intermittent()

in treeherder/push_health/classification.py [0:0]

• 25 lines of code
• 16 McCabe index (conditional complexity)

def set_intermittent(failure, previous_failures):
    # Not clear if we need these TODO items or not:
    # TODO: if there is >1 failure for platforms/config, increase pct
    # TODO: if >1 failures in the same dir or platform, increase pct

    name = failure["testName"]
    platform = failure["platform"]
    config = failure["config"]
    job_name = failure["jobName"]

    confidence = 0
    if name in previous_failures:
        confidence = 50
        if platform in previous_failures[name]:
            confidence = 75
            if config in previous_failures[name][platform]:
                confidence = 100

    # TODO: how many unique regression in win7*reftest*
    # Marking all win7 reftest failures as int, too many font issues
    if (
        confidence == 0
        and platform == "windows7-32"
        and ("opt-reftest" in job_name or "debug-reftest" in job_name)
    ):
        confidence = 50

    if failure["isClassifiedIntermittent"]:
        confidence = 100

    if confidence:
        failure["confidence"] = confidence
        failure["suggestedClassification"] = "intermittent"
        return True
    return False