These changes include storing monsters on shutdown, storing dead
characters and monsters, etc. Also fleshed out the MonsterManager's
internals a bit.
(cherry picked from commit af982710f6de050c09ef503cf2f8e1c9fdd34979)
Although the room transitions are working fairly well, it is still heavy
handed, and a number of optimizations can be done. On the whole, this
needs a review.
I've hacked the trigger system to create a really basic teleport pad,
using the dirt tile as an indicator. This behaviour is not coded into the
engine, but is in fact scripted in lua.
This commit is messy, due to lack of sleep.
There was a bug in the entity API, where getMotion() was pushing the
origin instead of the motion. This has been corrected. Since this is an
important bug, and because the features for this leg is finished, I'll
merge this to master.
Instead of using a massive block of case statements in serial_utility.cpp,
I've added FORMAT_* tags to SerialPacketType as a way to destinguish
between type values, at least internally. I can't believe I missed this
for so long.
I've also added a placeholder for the network API, as I was working on
that when I ran into this problem.
I need to devise a way to update the clients about changes to their
characters directly from the lua scripts. This isn't too important per se,
but the pattern will be important for the monster and trigger systems.
It seems like the project as a whole is fairly stable now. I'm prepping to
merge this into master, despite the lack of monsters ATM. Hopefully this
break hasn't affected the stability too much.
This new method rolls three otherwise similar methods together. There is
still a conditional which handles room movements separately, but it's much
smoother, and PumpPacketProximity is utilized much more.
I've also added a stub for graphical attack data via
HandleCharacterAttack()
Kayne Ruse