# lateral A clone of the command line utility by akramer. I wanted an excuse to write some Rust and am not creative enough to come up with my own ideas so I just reimplement other people's good ideas. A simple process parallelizer to make lives better at the commandline. Have you ever used `xargs -P -n 1` in an attempt to run slow commands in parallel? Gotten burned or gave up? Forgot `-0` when you needed it? Had to use positional arguments? Wanted to substitute text into the commandline? Lateral is here to help. ## Usage lateral 0.1.0 A simple process parallelizer to make lives better at the commandline. USAGE: lateral [OPTIONS] FLAGS: -h, --help Prints help information -V, --version Prints version information OPTIONS: -s, --socket UNIX domain socket path (default $HOME/.lateral/socket.$SESSIONID) SUBCOMMANDS: config Change the server configuration getpid Print pid of server to stdout help Prints this message or the help of the given subcommand(s) kill Kill the server with fire run Run the given command in the lateral server start Start the lateral background server wait Wait for all currently inserted tasks to finish ## Examples Example usage: lateral start for i in $(cat /tmp/names); do lateral run -- some_command $i done lateral wait With comments: # start a lateral server - one per session (login shell), runs 10 parallel tasks by default lateral start for i in $(gather list of work); do # Hand off the command to lateral lateral run -- my_slow_command $i done # Wait for all the work to be done lateral wait # wait stops the server when all commands are complete # if any commands returned a non-zero status, wait returns non-zero echo $? This is the most basic usage, and it's simpler than using xargs. It also supports much more powerful things. It doesn't only support command-line arguments like xargs, it also supports filedescriptors: lateral start for i in $(seq 1 100); do lateral run -- my_slow_command < workfile$i > /tmp/logfile$i done lateral wait The stdin, stdout, and stderr of the command to be run are passed to lateral, and so redirection to files works. This makes it trivial to have per-task log files. The parallelism is also dynamically adjustable at run-time. lateral start -p 0 # yup, it will just queue tasks with 0 parallelism for i in $(seq 1 100); do lateral run -- command_still_outputs_to_stdout_but_wont_spam inputfile$i done lateral config -p 10; lateral wait # command output spam can commence This also allows you to raise parallelism when things are going slower than you want. Underestimate how much work your machine can do at once? Ratchet up the number of tasks with `lateral config -p `. Turns out that you want to run fewer? Reducing the parallelism works as well - no new tasks will be started until the number running is under the limit. ## How does this black magic work? `lateral start` starts a server that listens on a unix socket that (by default) is based on your session ID. Each invocation of `lateral` in a different login shell will be independent. Unix sockets support more than passing data: they also support passing filedescriptors. When `lateral run` is called, it sends the command-line, environment, and open filedescriptors over the unix socket to the server where they are stored and queued to be run.