Since the last post, this is what's been done:
- Fixed some rogue bugs in the type conversion code
- Made naming of types a bit more consistent
- Various improvements to the parser
- Eliminated unused data fields and records in the binary format
- Binary loader is now complete
- Fixed a bug in Exegen that would output the wrong file names
- Fixed some type safety issues with structures that contain function references
- Implemented short-circuiting for the and() and or() functions
- Fixed serialization of several operations
- Fixed some assorted bugs in the binary loader
So definitely a productive day overall, but I'm still going to hack away into the night and see what else I can whip up.
My goal is to get a bit more polishing and cleanup finished, then start working on the multiprocessing support. Since that represents the largest unknown, I want to make sure I have plenty of time before the GDC deadline to get it done. Any extra time left is just a bonus; I haven't decided what to do with that time yet, if it even ends up happening.
There's still a couple of niggling annoyances with nested structure support (like the idiotically long chains of member(member(member(struct, field), field2), field3)) which I will try to attack if I get the time. Other than that, the only thing that really has to be done is some documentation and prep work for the demo kit.