关闭流/套接字和try-catch-finally

这是一个例子我有问题（它来自另一个SO问题）：

Here is an example I have questions about (it comes from another SO question):

public static void writeToFile (final String filename)
{
    PrintWriter out = null;     
    FileOutputStream fos = null;

    try
    {
        fos = new FileOutputStream(filename);
        out = new PrintWriter(new BufferedWriter(
                              new OutputStreamWriter(fos, "UTF-8")));

        for (final String line : log)
        {
            out.println(line);
        }

        out.flush();
        out.close();
    }
    catch (final Exception e)
    {
        System.err.println("Unable to write log to file.");
    }
    finally
    {
        if (fos != null)
        {
            try
            {
                fos.close();
            }
            catch (final IOException e)
            {
                System.err.println("Unable to write log to file.");
            }
        }
    }
}

我认为这个代码工作正常，并释放所有的资源，它应该等等。我的问题是：

Now I think this code works fine and releases all the resources where it should etc. My questions are:

1. try-catch-finally $ c $>中最后部分中的 FileOutputStream c>？当然，我可以把代码依次放在 try-catch ？

1. Why should I be closing the FileOutputStream in the finally section of the try-catch-finally? Surely I could just put the code sequentially after the try-catch?

为什么必须关闭 FileOutputStream ，而不是简单地替换 new OutputStreamWriter（fos，... 新的FileOutputStream（filename），... ？如果我先关闭FileOutputStream，会自动关闭其余和释放资源吗？同样的问题适用于套接字，如果我关闭一个套接字连接自动关闭流读取器/写入器并释放资源？

Why must I close the FileOutputStream separately rather than simply replacing new OutputStreamWriter(fos, ... with new OutputStreamWriter(new FileOutputStream(filename), ...? If I close the FileOutputStream first, will that automatically close the rest and release the resources? Same questions apply for sockets, if I close a socket connection does that automatically close the stream readers/writers and release the resources?

我已经反复告知，以确保我用UTF-到不同的系统具有不同的字符集（或某事沿着这些线）。这是仍然适用于读/写RAW字节数据（例如从非文本文件，或加密的结果），因为我认为charsets在哪里做文本只有字符？

I've been told repeatedly to ensure that I read and write streams with "UTF-8" due to different systems having different charsets (or something along those lines). Is this still applicable when reading/writing RAW byte data (say from a non-text file, or the result of an encryption) as I thought charsets where to do with textual characters only?