I am currently working hard making bugs testable, ironing bumps out and enhance the use of classpaths in general.

It unfortunately seems to be a huge effort to make the linter-javac-plugin 🚬-testable without touching the plugin itself. I currently don't want to add any untested functionality to prevent delivering a failure to our users.

Therefore i file this issue to collect ideas and feedback on the rewrite which will be accomplished with the "Release 2.0.0"-Milestone.

🎅 Wish List

Every wish will become a separate issue, to allow us the detailed discussion of concerns.

• Make the specs as nice as at linter-php. ✨
• Use the sourcepath-option for speedup (see Linter javac painfully slow #98). 🐎
• Lint on edit! 🐎
• Separate concerns (put supportive logic into separate modules). ✂️
• Lower startup-time under 50ms. 🐎
• Always default to language en with UTF-8 encoding by using -J-Duser.language=en and -J-Dfile.encoding=UTF-8 (see Does not lint #36).
• Introduce the atom-javaenv-package to use its API to the Java-Environment. 🌟 🔧 🌟
• Cross-dependent Java-files should be considered (as in fixes #31. send all java files in the project to javac, so it has the… #32).
• Deliver javac-args by an argsfile not by commandline - see Cant figure out the proper configuration #58 for further information.
• Allow any path-config-settings to either contain path-names or a file-reference which then contains the path-settings (see Use -sourcepath to improve compilation speed #99). 💅
• Document deviously source-parts (like sb/execs cwd-argument has been dropped #100). 🔦
• Create a Wiki for Contributors. 💝
• Add Debug-mode/button/switch allowing users easily discovering misbehavior. 🐛 🔨
• Add metrics as @Arcanemagus mentioned

Help me

• finding ideas
• give feedback on the codebase
• create PR's