DaemonConductor

Run a script many times, compile it once.

DaemonConductor acts as drop-in replacement for a subset of julia.

~$ juliaclient --help

    juliaclient [switches] -- [programfile] [args...]

Switches (a '*' marks the default value, if applicable):

 -v, --version              Display version information
 -h, --help                 Print this message
 --project[=<dir>|@.]       Set <dir> as the home project/environment
 -e, --eval <expr>          Evaluate <expr>
 -E, --print <expr>         Evaluate <expr> and display the result
 -L, --load <file>          Load <file> immediately on all processors
 -i                         Interactive mode; REPL runs and `isinteractive()` is true
 --banner={yes|no|auto*}    Enable or disable startup banner
 --color={yes|no|auto*}     Enable or disable color text
 --history-file={yes*|no}   Load or save history

Currently the client makes use of unix sockets and =io_uring=, and so only works on Linux systems at the moment.

Usage

• Install somewhere (potentially in a separate package environment)
• Run using DaemonConductor; DaemonConductor.install()
• Use juliaclient as a (mostly) drop-in replacement for julia

Configuration

When the daemon starts, it pays attention to the following environmental variables:

• JULIA_DAEMON_SERVER (default: /run/user/$UID/julia-daemon.sock), the socket the client connects to.
• JULIA_DAEMON_WORKER_ARGS (default: --startup-file=no), arguments passed to the worker Julia processes (individual arguments are split on whitespace).
• JULIA_DAEMON_WORKER_MAXCLIENTS (default: 1), the maximum number of clients a worker may be attached to at once. Set to 0 to disable.
• JULIA_DAEMON_WORKER_EXECUTABLE (default: julia on PATH), the path to the Julia executable used by the workers.
• JULIA_DAEMON_WORKER_TTL (default: 7200, 2h), the number of seconds a worker should be kept alive for after the last client disconnects from it. This variable can be updated within the worker itself.

Similarly, the client pays attention to JULIA_DAEMON_SERVER to make sure it connects to the right socket.

Caveats

• A worker is started per-project, and so running the same thing in multiple projects will incur the worker startup and compile cost multiple times.
• Workers are currently kept alive forever, leading to ballooning memory usage. This should be addressed in the future.
• The REPL seems a little funky, and wasn’t designed to allow for multiple REPLs per Julia process.
• Sometimes the socket file disappears, for mysterious reasons.
• This is currently the result of a weekend of work, there are likely some minor issues that haven’t been shaken out yet.