Allow machine-readable stdout to enable parsing by diagnostics tools

[zenibako]

I use Neovim as my IDE and the none-ls plugin to generate diagnostics for my Apex and JS files; this enables me to embed issues detected in pmd and eslint into my active buffer so I can easily fix my code.

Ideally, I'd like to remove both pmd and eslint from my workflow and manage everything with code-analyzer commands. This would allow me to take advantage of the new yaml-based configuration file to maintain rules in my codebase.

none-ls allows me to write my own parser for the diagnostics output of code-analyzer run, but currently I am hitting a roadblock because there is no way to simply output the results in JSON format direct to stdout. I could definitely hack together a bash script to do this, but I have to force the code-analyzer run output to be quiet with > /dev/null 2>&1 and do several other operations to print then remove a .json output file. It seems ideal to simply output a --json and handle it that way.

Even better would be a drop-in replacement for pmd or eslint results, so I could just substitute those commands with sf code-analyzer run and have its output already interpreted by existing tools built to read those result formats.

Thoughts on this being desirable functionality for others as well?

[stephen-carter-at-sf]

Will take this under consideration now that we have gone GA with code-analyzer (v5).