使用 Java 读取/写入 linux 管道
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/1632069/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
Read/Write to linux Pipe using Java
提问by amit.bhayani
My query is on what is the best way to read / write to a linux Pipe in Java? I already am using the java.io.RandomAccessFile like
我的问题是在 Java 中读取/写入 linux Pipe 的最佳方式是什么？我已经在使用 java.io.RandomAccessFile 之类的
RandomAccessFile file = new RandomAccessFile("/dev/zap/16", "rw");
and then passing it to worker thread which reads it after every 2ms as
然后将其传递给工作线程，工作线程每 2 毫秒读取一次
byte[] buffer = new byte[16];
file.read(buffer);
It does read it from Pipe, but I suspect that some bytes are overwritten. Do you know how linux (ubuntu) handles the buffer for pipe's?
它确实从 Pipe 中读取它，但我怀疑某些字节被覆盖了。你知道 linux (ubuntu) 如何处理管道的缓冲区吗？
回答by sfussenegger
I haven't ever tried this myself but what your doing feels just wrong. Linux pipes are First in - First out (FIFO) by definition. Hence you should only be able to read bytes in the same order as you've written them - not randomly. I'd suggest to use a normal Fileinstead and it should work fine.
我自己从来没有尝试过这个，但你所做的感觉是错误的。Linux 管道根据定义是先进先出 (FIFO)。因此，您应该只能以与写入它们相同的顺序读取字节 - 而不是随机读取。我建议改用普通的File，它应该可以正常工作。
回答by Joachim Sauer
Pipes aren't handled in any way special by Java as far as I know. You simply open the file for writing and write to it.
据我所知，Java 并没有以任何特殊的方式处理管道。您只需打开文件进行写入并写入即可。
You can't really "overwrite" anything in a pipe, since you can't seek in a pipe. For the same reason a RandomAccessFileisn't the smartest choice to use (since a pipe is explicitely nota random access file). I'd suggest using a FileOutputStreaminstead.
你不能真正“覆盖”管道中的任何东西，因为你不能在管道中寻找。出于同样的原因， aRandomAccessFile不是最明智的选择（因为管道显然不是随机访问文件）。我建议使用 aFileOutputStream代替。
Also note that read()is not guaranteed to read until the buffer is full! It can read a single byte as well and you need to check its return value and possibly loop to read the full buffer.
另请注意，read()在缓冲区已满之前不能保证读取！它也可以读取单个字节，您需要检查其返回值并可能循环读取完整的缓冲区。
回答by Anurag Uniyal
I think you may not be flushing after writing, so do OutputStream.flush() often and read may be a byte at time, at least to see if your data is getting thru. e.g. to start with open a named pipe in readonly mode(FileInputStream) in process1, open it in write mode(FileOutputStream ) in process2, so anything you write in process2 will be read in process1.
我认为您在写入后可能不会刷新，因此经常使用 OutputStream.flush() 并且读取可能是一个字节，至少可以查看您的数据是否通过。例如，首先在 process1 中以只读模式（FileInputStream）打开命名管道，在 process2 中以写入模式（FileOutputStream）打开它，因此您在 process2 中写入的任何内容都将在 process1 中读取。
also read
还读
http://www.tldp.org/LDP/lpg/node15.html
http://www.unixguide.net/unix/programming/2.10.5.shtmlhttp://www.unixguide.net/unix/programming/2.10.6.shtml
http://www.tldp.org/LDP/lpg/node15.html
http://www.unixguide.net/unix/programming/2.10.5.shtml http://www.unixguide.net/unix/programming/ 2.10.6.shtml
