Piotr KrzemiĆski
11/14/2023, 7:31 AMilya.chernikov
11/15/2023, 9:52 AMPiotr KrzemiĆski
11/15/2023, 9:55 AMPiotr KrzemiĆski
11/15/2023, 9:55 AMilya.chernikov
11/15/2023, 10:05 AMilya.chernikov
11/15/2023, 10:05 AMPiotr KrzemiĆski
11/15/2023, 10:06 AMilya.chernikov
11/15/2023, 10:12 AMPiotr KrzemiĆski
11/15/2023, 12:01 PM@file:Import()
annotations and follow them. Once done, we can either inline the source code of the imported scripts so that the cache can calculate the hash using all files including imported, or have an object that will contain references to the imported files and build the hash iteratively using all files. After this step, proceed to regular compilation.
Does it make any sense?ilya.chernikov
11/15/2023, 1:10 PMwithScriptCompilationCache
down the callstack, around the end of ScriptJvmCompilerImplsKt.compileImpl
, where the actual compilation is called with already fully refined configurations.
(We'll need to extend withScriptCompilationCache
to accept multiple sources and configurations too. Theoretically, partial caching should be also possible at this place, but we can postpone it for now.)
This approach would also handle a potential problem with clearing the .m2
cache.