Github user thvasilo commented on a diff in the pull request:
https://github.com/apache/flink/pull/949#discussion_r36958345
 Diff: flinkcore/src/main/java/org/apache/flink/api/common/operators/util/BernoulliSampler.java

@@ 0,0 +1,107 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.flink.api.common.operators.util;
+
+import com.google.common.base.Preconditions;
+
+import java.util.Iterator;
+import java.util.Random;
+
+/**
+ * A sampler implementation built upon Bernoulli Trail. For sample with fraction and
without replacement,
+ * each element sample choice is just a bernoulli trail.
+ *
+ * @param <T> The type of sample.
+ */
+public class BernoulliSampler<T> extends RandomSampler<T> {
+
+ private final double fraction;
+ private final Random random;
+
+ /**
+ * Create a bernoulli sampler sample fraction and default random number generator.
+ *
+ * @param fraction Sample fraction, aka the bernoulli sampler possibility.
+ */
+ public BernoulliSampler(double fraction) {
+ this(fraction, new Random());
+ }
+
+ /**
+ * Create a bernoulli sampler sample fraction and random number generator seed.
+ *
+ * @param fraction Sample fraction, aka the bernoulli sampler possibility.
+ * @param seed Random number generator seed.
+ */
+ public BernoulliSampler(double fraction, long seed) {
+ this(fraction, new Random(seed));
+ }
+
+ /**
+ * Create a bernoulli sampler sample fraction and random number generator.
+ *
+ * @param fraction Sample fraction, aka the bernoulli sampler possibility.
+ * @param random The random number generator.
+ */
+ public BernoulliSampler(double fraction, Random random) {
+ Preconditions.checkArgument(fraction >= 0 && fraction <= 1.0d, "fraction
fraction must between [0, 1].");
+ this.fraction = fraction;
+ this.random = random;
+ }
+
+ /**
+ * Sample the input elements, for each input element, take a Bernoulli Trail for sample.
+ *
+ * @param input Elements to be sampled.
+ * @return The sampled result which is lazy computed upon input elements.
+ */
+ @Override
+ public Iterator<T> sample(final Iterator<T> input) {
+ if (fraction == 0) {
+ return EMPTY_ITERABLE;
+ }
+
+ return new SampledIterator<T>() {
+ T current;
+
+ @Override
+ public boolean hasNext() {
+ if (current == null) {
+ while (input.hasNext()) {
+ T element = input.next();
+ if (random.nextDouble() <= fraction) {
+ current = element;
+ return true;
+ }
+ }
+ current = null;
+ return false;
+ } else {
+ return true;
+ }
+ }
+
+ @Override
+ public T next() {
 End diff 
It feels a bit counterintuitive that the next element is prepared in the `hasNext()` function.
Doesn't this mean that `hasNext()` **needs** to be called every time before we call `next()`?
Can we protect against that case where we would get a `null` element back that way?

If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.

