core/src/main/java/org/apache/brooklyn/enricher/stock/YamlRollingTimeWindowMeanEnricher.java [154:186]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        long windowStart = Math.max(now-timePeriod.toMilliseconds(), firstTimestamp);
        long windowEnd = Math.max(now-timePeriod.toMilliseconds(), lastTimestamp);
        Double confidence = ((double)(windowEnd - windowStart)) / timePeriod.toMilliseconds();
        if (confidence <= 0.0000001d) {
            // not enough timestamps in window 
            double lastValue = values.get(values.size()-1).doubleValue();
            return lastAverage = new ConfidenceQualifiedNumber(lastValue, 0.0d);
        }
        
        long start = windowStart;
        long end;
        double weightedAverage = 0.0d;
        
        Iterator<T> valuesIter = values.iterator();
        Iterator<Long> timestampsIter = timestamps.iterator();
        while (valuesIter.hasNext()) {
            // Ignores null and out-of-date values (and also values that are received out-of-order, but that shouldn't happen!)
            Number val = valuesIter.next();
            Long timestamp = timestampsIter.next();
            if (val!=null && timestamp >= start) {
                end = timestamp;
                weightedAverage += ((end - start) / (confidence * timePeriod.toMilliseconds())) * val.doubleValue();
                start = timestamp;
            }
        }
        
        return lastAverage = new ConfidenceQualifiedNumber(weightedAverage, confidence);
    }
    
    /**
     * Discards out-of-date values, but keeps at least one value.
     */
    private void pruneValues(long now) {
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



policy/src/main/java/org/apache/brooklyn/policy/enricher/RollingTimeWindowMeanEnricher.java [167:199]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        long windowStart = Math.max(now-timePeriod.toMilliseconds(), firstTimestamp);
        long windowEnd = Math.max(now-timePeriod.toMilliseconds(), lastTimestamp);
        Double confidence = ((double)(windowEnd - windowStart)) / timePeriod.toMilliseconds();
        if (confidence <= 0.0000001d) {
            // not enough timestamps in window 
            double lastValue = values.get(values.size()-1).doubleValue();
            return lastAverage = new ConfidenceQualifiedNumber(lastValue, 0.0d);
        }
        
        long start = windowStart;
        long end;
        double weightedAverage = 0.0d;
        
        Iterator<T> valuesIter = values.iterator();
        Iterator<Long> timestampsIter = timestamps.iterator();
        while (valuesIter.hasNext()) {
            // Ignores null and out-of-date values (and also values that are received out-of-order, but that shouldn't happen!)
            Number val = valuesIter.next();
            Long timestamp = timestampsIter.next();
            if (val!=null && timestamp >= start) {
                end = timestamp;
                weightedAverage += ((end - start) / (confidence * timePeriod.toMilliseconds())) * val.doubleValue();
                start = timestamp;
            }
        }
        
        return lastAverage = new ConfidenceQualifiedNumber(weightedAverage, confidence);
    }
    
    /**
     * Discards out-of-date values, but keeps at least one value.
     */
    private void pruneValues(long now) {
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -