有个功能需要同时上传N个文件。代码如下：

ApiService as = ApiManager.getApiService();

final ExecutorService es = Executors.newFixedThreadPool(9);

final int count = Bimp.tempSelectBitmap.size();

final CountDownLatch finishedLatch = new CountDownLatch(count);

final long start = System.currentTimeMillis();

for (int k = 0; k < count; k++) {

final String fp = Bimp.tempSelectBitmap.get(k).getImagePath();

RequestBody fbody = RequestBody.create(MediaType.parse("image/*"), new File(fp));

as.uploadAttach(fbody)

.subscribeOn(Schedulers.from(es))

.observeOn(Schedulers.computation())

.subscribe(new Subscriber<UploadAttachJSON>() {

@Override

public void onCompleted() {

}

@Override

public void onError(Throwable e) {

finishedLatch.countDown();

Log.e("UPLOAD FAILED -------->", fp);

}

@Override

public void onNext(UploadAttachJSON uploadAttachJSON) {

finishedLatch.countDown();

sb.append(uploadAttachJSON.url).append(",");

Log.e("UPLOADED IMAGE URL -->", uploadAttachJSON.url);

h.post(new Runnable() {

@Override

public void run() {

pd.setMessage("正在上传...  " + (count - finishedLatch.getCount()) + "/" + count);

}

});

}

});

}

try {

finishedLatch.await();

} catch (InterruptedException e) {

e.printStackTrace();

}

long end = System.currentTimeMillis();

Log.e("IMAGE UPLOAD COMPLETED", (end - start) + "");

es.shutdown();

以上为并行的写法。从线程池中拿出N个线程来同时上传这N个文件。

串行写法：

.subscribeOn(Schedulers.io())

或者 用

Observable.merge

来合并这些请求。

结果发现并行和串行所花费的时间几乎都差不多。。 是不是和android底层有关？这些网络请求其实最后都被底层给block了，然后串行出去？

设备的网速是不是有限制