0

Iterating on my compiler plugin's code, I am publishing my compiler plugin to my local ivy repository after each compilation of it (via publishLocal), and then running my other project where a dependency upon this plugin is defined via addCompilerPlugin. Is there a more concise practice for developing a compiler plugin?

Of course, I could aggregate the two into a multi-project build definition. But it might be nice to learn of more lightweight practices for iterating plugin code...

Could I in the very least depend on the compiler plugin without turning it into a library for that? from the syntax permitted by addCompilerPlugin it looks like a library must (?) be created and added, rather than affording a dependency on mere class files.

matanster
  • 15,072
  • 19
  • 88
  • 167

1 Answers1

2

Look at what I do in the scapegoat plugin, where I create a 'test' compiler. I use this to compile code snippets in the form of unit tests.

This way you can write code and run your tests, as you would normally, without needing to publish externally.

https://github.com/sksamuel/scalac-scapegoat-plugin/blob/master/src/test/scala/com/sksamuel/scapegoat/PluginRunner.scala

sksamuel
  • 16,154
  • 8
  • 60
  • 108