java - Java流图中可重用的单实例包装器/对象

Question

似乎这个问题应该已经有了答案，但我找不到重复的答案。
无论如何，我想知道社区对Stream.map用例有何看法？

Wrapper wrapper = new Wrapper();
list.stream()
    .map( s -> {
        wrapper.setSource(s);
        return wrapper;
    } )
    .forEach( w -> processWrapper(w) );
    
public static class Source {
    private final String name;
        
    public Source(String name) {
        this.name = name;
    }
        
    public String getName() {
        return name;
    }
}
    
public static class Wrapper {
    private Source source = null;
        
    public void setSource(Source source) {
        this.source = source;
    }
        
    public String getName() {
        return source.getName();
    }
}

public void processWrapper(Wrapper wrapper) {
}

我不是 map用法的忠实拥护者，但是在处理大型流时，它可能有助于提高性能，并避免为每个 Wrapper创建不必要的 Source。
这肯定有其局限性，例如对于并行流和像 collect这样的终端操作几乎是无用的。
更新-
问题不是关于“怎么做”，而是“我可以这样吗”。例如，我可以有一个仅适用于Wrapper的代码，并且我想在 forEach中调用它，但是想要避免为每个 Source元素创建一个新实例。
基准测试结果
显示有关 的8倍的可折叠包装器改善的折叠效果-

Benchmark (N) Mode Cnt Score Error Units

BenchmarkTest.noReuse 10000000 avgt 5 870.253 ± 122.495 ms/op

BenchmarkTest.withReuse 10000000 avgt 5 113.694 ± 2.528 ms/op

基准代码-

import java.util.List;
import java.util.ArrayList;
import java.util.concurrent.TimeUnit;

import org.openjdk.jmh.annotations.*;
import org.openjdk.jmh.infra.Blackhole;
import org.openjdk.jmh.runner.Runner;
import org.openjdk.jmh.runner.options.Options;
import org.openjdk.jmh.runner.options.OptionsBuilder;

@BenchmarkMode(Mode.AverageTime)
@OutputTimeUnit(TimeUnit.MILLISECONDS)
@State(Scope.Benchmark)
@Fork(value = 2, jvmArgs = {"-Xms2G", "-Xmx2G"})
public class BenchmarkTest {

    @Param({"10000000"})
    private int N;

    private List<Source> data;

    public static void main(String[] args) throws Exception {
        Options opt = new OptionsBuilder()
            .include(BenchmarkTest.class.getSimpleName())
            .forks(1)
            .build();
        new Runner(opt).run();
    }

    @Setup
    public void setup() {
        data = createData();
    }

    @Benchmark
    public void noReuse(Blackhole bh) {
        data.stream()
            .map( s -> new Wrapper1( s.getName() ) )
            .forEach( t -> processTarget(bh, t) );
    }

    @Benchmark
    public void withReuse(Blackhole bh) {
        Wrapper2 wrapper = new Wrapper2();
        data.stream()
            .map( s -> { wrapper.setSource(s); return wrapper; } )
            .forEach( w -> processTarget(bh, w) );
    }
    
    public void processTarget(Blackhole bh, Wrapper t) {
        bh.consume(t);
    }
    
    private List<Source> createData() {
        List<Source> data = new ArrayList<>();
        for (int i = 0; i < N; i++) {
            data.add( new Source("Number : " + i) );
        }
        return data;
    }
    
    public static class Source {
        private final String name;

        public Source(String name) {
            this.name = name;
        }

        public String getName() {
            return name;
        }
    }

    public interface Wrapper {
        public String getName();
    }
    
    public static class Wrapper1 implements Wrapper {
        private final String name;

        public Wrapper1(String name) {
            this.name = name;
        }

        public String getName() {
            return name;
        }
    }
    
    public static class Wrapper2 implements Wrapper {
        private Source source = null;

        public void setSource(Source source) {
            this.source = source;
        }

        public String getName() {
            return source.getName();
        }
    }
}

完整基准测试报告-

# JMH version: 1.21
# VM version: JDK 1.8.0_191, Java HotSpot(TM) 64-Bit Server VM, 25.191-b12
# VM invoker: /Library/Java/JavaVirtualMachines/jdk1.8.0_191.jdk/Contents/Home/jre/bin/java
# VM options: -Xms2G -Xmx2G
# Warmup: 5 iterations, 10 s each
# Measurement: 5 iterations, 10 s each
# Timeout: 10 min per iteration
# Threads: 1 thread, will synchronize iterations
# Benchmark mode: Average time, time/op
# Benchmark: BenchmarkTest.noReuse
# Parameters: (N = 10000000)

# Run progress: 0.00% complete, ETA 00:03:20
# Fork: 1 of 1
# Warmup Iteration   1: 1083.656 ms/op
# Warmup Iteration   2: 846.485 ms/op
# Warmup Iteration   3: 901.164 ms/op
# Warmup Iteration   4: 849.659 ms/op
# Warmup Iteration   5: 903.805 ms/op
Iteration   1: 847.008 ms/op
Iteration   2: 895.800 ms/op
Iteration   3: 892.642 ms/op
Iteration   4: 825.901 ms/op
Iteration   5: 889.914 ms/op


Result "BenchmartTest.noReuse":
  870.253 ±(99.9%) 122.495 ms/op [Average]
  (min, avg, max) = (825.901, 870.253, 895.800), stdev = 31.812
  CI (99.9%): [747.758, 992.748] (assumes normal distribution)


# JMH version: 1.21
# VM version: JDK 1.8.0_191, Java HotSpot(TM) 64-Bit Server VM, 25.191-b12
# VM invoker: /Library/Java/JavaVirtualMachines/jdk1.8.0_191.jdk/Contents/Home/jre/bin/java
# VM options: -Xms2G -Xmx2G
# Warmup: 5 iterations, 10 s each
# Measurement: 5 iterations, 10 s each
# Timeout: 10 min per iteration
# Threads: 1 thread, will synchronize iterations
# Benchmark mode: Average time, time/op
# Benchmark: BenchmarkTest.withReuse
# Parameters: (N = 10000000)

# Run progress: 50.00% complete, ETA 00:01:58
# Fork: 1 of 1
# Warmup Iteration   1: 113.780 ms/op
# Warmup Iteration   2: 113.643 ms/op
# Warmup Iteration   3: 114.323 ms/op
# Warmup Iteration   4: 114.258 ms/op
# Warmup Iteration   5: 117.351 ms/op
Iteration   1: 114.526 ms/op
Iteration   2: 113.944 ms/op
Iteration   3: 113.943 ms/op
Iteration   4: 112.930 ms/op
Iteration   5: 113.124 ms/op


Result "BenchmarkTest.withReuse":
  113.694 ±(99.9%) 2.528 ms/op [Average]
  (min, avg, max) = (112.930, 113.694, 114.526), stdev = 0.657
  CI (99.9%): [111.165, 116.222] (assumes normal distribution)


# Run complete. Total time: 00:03:40

REMEMBER: The numbers below are just data. To gain reusable insights, you need to follow up on
why the numbers are the way they are. Use profilers (see -prof, -lprof), design factorial
experiments, perform baseline and negative tests that provide experimental control, make sure
the benchmarking environment is safe on JVM/OS/HW level, ask for reviews from the domain experts.
Do not assume the numbers tell you what you want them to tell.

Benchmark                     (N)  Mode  Cnt    Score     Error  Units
BenchmarkTest.noReuse    10000000  avgt    5  870.253 ± 122.495  ms/op
BenchmarkTest.withReuse  10000000  avgt    5  113.694 ±   2.528  ms/op

Answer 1

您可以使用一些方便的功能，也可以使用线程安全版本来并行处理。

Function<T,U> threadSafeReusableWrapper(Supplier<U> newWrapperInstanceFn, BiConsumer<U,T> wrapFn) {
   final ThreadLocal<T> wrapperStorage = ThreadLocal.withInitial(newWrapperInstanceFn);
   return item -> {
      T wrapper = wrapperStorage.get();
      wrapFn.consume(wrapper, item);
      return wrapper;
   }
}

Function<T,U> reusableWrapper(U wrapper, BiConsumer<U,T> wrapFn) {
   return item -> {
      wrapFn.consume(wrapper, item);
      return wrapper;
   };
}

list.stream()
    .map(reusableWrapper(new Wrapper(), Wrapper::setSource))
    .forEach( w -> processWrapper(w) );
list.stream()
    .map(threadSafeReusableWrapper(Wrapper::new, Wrapper::setSource))
     .parallel()
    .forEach( w -> processWrapper(w) );

但是，我认为这不值得。这些包装器生命周期短，因此不太可能离开年轻一代，因此会很快被垃圾收集。虽然，我认为这个想法值得与 micro-benchmark library JMH一起检查