Pass file descriptor through pipe

How can I watch this process's stdout? Improve this question. Pipe Pipe 11 2 2 bronze badges. Or just to find out where the stream is going?

I removed the embedded tag because this question doesn't seem to be about embedded systems. If any aspects of what you are asking related to a specific system, please edit your question and add any relevant information about the system you are on to it.

Add a comment. Active Oldest Votes. Improve this answer. Hauke Laging Hauke Laging But the goal of this answer was to illustrate redirections in the shell. Here's an example of using extra FDs as bash script chattiness control:! Fordi Fordi 4 4 silver badges 7 7 bronze badges. What's the purpose of using eval here? You have to use eval to make them work properly. I know this isn't a new answer, but I had to stare at this quite awhile to see what it does and thought it would be helpful if someone added an example of this function being used.

This one echos and captures the whole output of a command - df, in this case. Example: using flock to force scripts to run serially with file locks One example is to make use of file locking to force scripts to run serially system wide. Usage of lock and unlock functions in scripts You can use it in your scripts like the following example. Sam Gleske Sam Gleske 3 3 bronze badges.

Ben Blank Ben Blank 4 4 bronze badges. Additional file descriptors can be used for creating temporary files in shell scripts. Sign up or log in Sign up using Google. Sign up using Facebook. Sign up using Email and Password. Post as a guest Name.

Email Required, but never shown. The Overflow Blog. Podcast Helping communities build their own LTE networks. Podcast Making Agile work for data science. Stack Overflow for Teams — Collaborate and share knowledge with a private group. Create a free Team What is Teams? Collectives on Stack Overflow.

Learn more. Asked 10 years, 4 months ago. Active 2 years, 11 months ago. Viewed 2k times. Any help is appreciated. Improve this question. Francesco Boi 6, 9 9 gold badges 62 62 silver badges bronze badges. When we run commands in the terminal, any input and output need to be handled appropriately. The process that gets created for each command needs to know what data, if any, to take in as input and possibly what data to output.

The above figure represents the default setup of the input and output streams. I represent the flow of data from left to right. Put another way: input is read from somewhere; output is written somewhere. This mental model will prove helpful in future diagrams that are more complex. One thing to note is that there are actually two streams that can write output to the terminal by default: stdout and stderr.

The stderr stream is used when something goes wrong when trying to execute a command. In this example, the stream that is used to display the second line is actually stderr, not stdout. Since stderr also goes to the terminal by default, we see the error message in the terminal. Conceptual data flow for the standard input 0 , output 1 , and error 2 streams.

Just remember that it exists! Now we can explore data flow using commands. Some commands both read input and write output, but others only do one or neither. Command arguments that are options are really read in from the command line as an argument array ; actual input is read in from an open file that is associated with a file descriptor.

If a file is passed as an argument, then I consider it input if the process will actually read or manipulate the contents of that file e. As a side note, command line option arguments are the result of another Unix design choice that allows behavior modification of an executed command to be passed in separately from the received input. Keeping the arguments and input separate makes life easier when pipes are involved. To illustrate a command that can have no input but has output, consider ls , which lists all the files in the current directory:.

If I give it the name of a file or directory that can be moved or renamed successfully, then no data is output via stdout or stderr.

If, however, I use mv incorrectly such that an error occurs, then I will have output to stderr:. One of my favorite examples of a command that both reads input and writes output is sort. When used with no file arguments and no input redirection, the terminal waits for the user to enter the strings to sort one string per line.

Once the user types Ctrl-D which closes the write end of the communication channel that connects the keyboard to the stdin of the sort process , the process running sort will know that all desired strings have been entered. Thus these strings are passed via stdin into the process that runs the command, sorted by said process, and then written to the terminal via stdout.

Pretty nifty! The bolded strings are user input and the strings that follow represent the sorted output. The sort command. Input is typed into the keyboard, then the output is displayed in sorted order. Note that sort can also take a filename argument to get input from the specified file instead of waiting for data to be entered by the user for example, sort words. Now that we understand the general idea of data flow from stdin to stdout or stderr, we can discuss how to control the flow of input and output.

Fun stuff! Unix has a simple yet valuable design philosophy, as explained by Doug McIlroy, the inventor of the Unix pipe:. Write programs to work together.

Write programs to handle text streams, because that is a universal interface. The concept of a pipe is extremely powerful. Pipes allow data from one process to be passed to another via unidirectional data flow so that commands can be chained together by their streams.

This allows commands to work together to achieve a larger goal. This chaining of processes can be represented by a pipeline : commands in a pipeline are connected via pipes, where data is shared between processes by flowing from one end of the pipe to the other. Since each command in the pipeline is run in a separate process, each with a separate memory space, we need a way to allow those processes to communicate with each other.

This is is exactly the behavior that the pipe system call provides. Implementation-wise, pipes are actually just buffered streams that are associated with two file descriptors that are set up so that the first one can read in data that is written to the second one. Specifically, in the code written to handle the execution of commands in a pipeline, an array of two integers is created and a pipe call populates the array with two available file descriptors usually the lowest two values available such that the first file descriptor in the array can read in data written to the second.

Physical pipes are naturally a great analogy for this abstraction. We can think of the data stream that starts in one process as water in an isolated environment, and the only way to allow the water to flow to the environment of the next process is to connect the environments with a pipe.