I would like to be able to do the following:
- Write and compile a C extension containing nothing but singleton
methods defined on “self”. - Load the extension multiple times, attaching these methods to
different objects. - Make changes to the extension, rebuild it, and replace the defined
methods on select objects.
#1 is easy, #2 can be done by linking multiple singletons to the same
method.
But #3 cannot be done, because I can’t dlclose() extensions.
Why not?
I know I could just create new instances of the interpreter.
I don’t want to do this - I want to be able to create a unified,
persistent object model that avoids the hassle of IPC, but still has the
ability to load and unload dynamically compiled machine code.
Do I need to create my own extension loading/unloading library? Or is
there an obscure feature or upstream patch I’m not aware of?