[Android]Why can we still use setOutputFile(String path) of MediaRecorder?

Question

I am trying to extract the muxer part of MediaRecorder, so I am learning the source code of Android 4.2.2 now, anyway, that is not important. I have a questions here. It's about setOutputFile(String path) of MediaRecorder.java.

The call consequence of MediaRecorder.java is

MediaRecorder.java
  +- android_media_MediaRecorder.cpp
       +- MediaRecorder.cpp
            +- MediaPlayerService.cpp
                 +- MediaRecorderClient.cpp
                      +- StagefrightRecorder.cpp

But in the StagefrightRecorder.cpp, it says

status_t StagefrightRecorder::setOutputFile(const char *path) {
    ALOGE("setOutputFile(const char*) must not be called");
    // We don't actually support this at all, as the media_server process
    // no longer has permissions to create files.

    return -EPERM;
}

Then how can we continue to use setOutputFile(String path) of MediaRecorder? I really don't understand.

Answer 1

In MediaRecorder.java, setOutputFile is handled at the Java layer itself without a corresponding native implementation. In the Java implementation, the path is stored into mPath.

When prepare is called, a FileOutputStream is created and passed to the native implementation of setOutputFile which takes in a File Descriptor as can be observed from this implementation. This method i.e. taking in a File Descriptor is supported by the native implementation and hence, the output is generated and stored into the file system.

Hence, in examples such as MediaRecorderTest.java, one can observe that MediaRecorder object is initialized with a string for the output file path.