Currently ScalaLoader scans all classes in a plugin's jar file for special annotations from which it detects the plugin's main class, and to detect which bytecode transformations to apply and where. This happens every time a ScalaPlugin loads.
This could be optimized by caching the scan results in a file on disk (and in-memory even), and if the hash of (plugin + ScalaLoader version) hasn't changed on reload, then those scan results from a previous run can be used directly.
Currently ScalaLoader scans all classes in a plugin's jar file for special annotations from which it detects the plugin's main class, and to detect which bytecode transformations to apply and where. This happens every time a ScalaPlugin loads.
This could be optimized by caching the scan results in a file on disk (and in-memory even), and if the hash of (plugin + ScalaLoader version) hasn't changed on reload, then those scan results from a previous run can be used directly.