jaredgorski/arpx

Small-scale process orchestration

arpx's Issues

Fix thread blocking/unblocking behavior or remove broken feature

Add main thread blocking for processes and actions

Move logic in mod files to dedicated logic files; use mod files for module exports

Reconfigure profiles and config

Allow passing a profile via the command by default and use the config for customizing system settings

Add filesystem triggers

Fix out of order events

Replace log triggers concept with logical condition statement

Instead of being limited to includes_string and includes_regex, users could supply any condition which, when executed upon logging the current stdout/stderr line, exits with success or failure (denoting true/false condition status)

Allow thread blocking actions to resume/unblock thread when finished

Add blocking option on actions for synchronous behavior

Fix incorrect README asset

^^ loop2 and loop3 are mixed up in the description

Add to homebrew

https://docs.brew.sh/Formula-Cookbook

Add graceful shutdown (intercept ctrl-c)

Fix clippy warnings

Add doc comments

https://doc.rust-lang.org/stable/rust-by-example/meta/doc.html

Backlog

Backlog

This issue documents the backlog which serves as the basis for future versions. If you have opinions on this, feel free to comment and start discussions. Some of these ideas may be implemented and your input matters.

Features

• Make bin configuration more streamlined:
  • Presets?
    • bash: sh -c
    • cmd: cmd /c
    • powershell: powershell -Command
  • Enable configuring bin via profile?
• ? External log monitors:
  • use arpx to monitor external log file and spawn processes ontrigger
• ?? Per-process log colors:
  • probably not for Arpx... more appropriate for a separate program ("prism"?)

Changes

• Log monitors should be able to stdin process threads if needed
• Consider renaming test property in log_monitors mappings to command, to be consistent with processes
• Consider renaming command to exec everywhere
• Consider renaming ontrigger property in log_monitors mappings to onsucceed, or assign it to a special exit code instead (such as 2) and call it on2
• Allow log monitors to execute multiple (concurrent) ontrigger actions
• Log monitors should spawn ontrigger actions on the process threads they watch (not on their own thread)
  • park process threads if needed and use message passing to prompt the task thread to unpark and run the ontrigger
• Consider making jobs composable (jobs can be actions too):
  • jobs would now have contingency
  • runtime ctx would need a job_map
  • thread max would be harder to enforce

Improvements

• Insofar as ctx is passed around, the entire log_monitor_map and process_map objects get copied. Can ctx be referenced instead?

Platform support

• Windows support pending this issue

Documentation

• Replace foobar examples throughout docs (and demo) with standard, practical examples (cluster init, webserver, api service, etc.)
• Migration guide for profiles
• Clarify how log monitors should be used

Add ability for shell-type custom actions to execute stdin when invoked

Add tests

Add CI + binary release

Build out README

Make more scriptable

Add notifier actions

Update documentation

• Use rustdoc
• Actions stdin option is not documented
• etc.

Split into arpx_core and arpx_cli crates

Defend against reaching thread max on current OS

Improve errors

https://docs.rs/failure/0.1.6/failure/
https://docs.rs/failure-tools/4.0.5/failure_tools/

Rewrite README

README in its current form is confusing. Some needs include: better differentiation between processes and actions (primary processes vs side effects), better examples (including calling processes onsucceed/onfail), less all-caps... it's obnoxious, and more.

Probably best to just rewrite and organize the concepts more simply.

Add browser actions

Add status triggers (onfail and onsucceed)