@k4t3 on that last point: nah to the package manager (it doesn’t need one). also i’ve implemented compression algorithms in luajit and they were not too slow to use, so imo JIT is good enough here. use a real language if you want speed.
but i can see how passing bytecode around might be useful in cases where you want that backwards-compatibility: you have a generic transform tool that will just output the data you select in a format that legacy software can handle, and that bytecode implements efficient access/transforms for the specific data structures your program uses
…or something like that
@mia@shrimptest.0x0.st