Proper pipeline
-
Right now, we optimise the ABC files separately and then call the bytecode generator on all optimised ABC files at once. In tests.txt
we store which modules are needed. For a usable system, this needs to be integrated withcpm
and perhapsclm
. -
The binary format is suitable for an interpreter but lacks information that is needed to build dynamics from it. We need to add a symbol table. This will in the future allow for - Generating bytecode for all modules separately and linking them together (which can be faster, because library modules don't have to be regenerated); - Stripping the bytecode (both after compilation and when building a dynamic) → will be done in #11.
Edited by Clean Importer