Remove Eluna* from being passed. You can now get it through Eluna::GetEluna(L)
Change function call cleanup to lua_settop instead of manual loop
Explicitly delete the copy constructor and copy assignment operators
Method documentation should be completed. Only thing left now is polishing up (fixing typos, update wording, add enumerator data to method documentation that doesn't have it, etc)..
Cleaned whitespaces
Corrected/Fixed up some comments
Removed all known documented method comments in LuaFunctions.cpp. Commented out methods are now marked as undocumented until they are uncommented & documented in their respective header files.
Removed 'Getter, Setter, Boolean, Other' comments in method files
Documented VehicleMethods, the rest of PlayerMethods, UnitMethods, CreatureMethods and GuildMethods.
Refer to #101 for more information
Some fixes for TC and changes overall
Pass map object to hook via function arguments
The map object is no longer stored inside the instance data table.
Fix mistake in base64 decoder
It was failing whenever it encountered a '=' character, which is
completely valid.
Make ElunaInstanceAI::Load always load something
When it failed to load data, it was leaving nothing on the stack. Since
subsequent code expected Load to always load something, this was causing
issues.
Now, when Load fails to load anything, it just leaves a new empty table on
the stack.
Also: the error messages for Load have been improved.
Modify lua-marshal to allow saving of functions/userdata.
Some additional code was needed to save functions due to the inclusion of
a reference to _ENV within their upvalues (since Lua 5.2).
During encoding, a placeholder is left where the _ENV reference would be.
During decoding, a reference to the current _G is swapped with the
placeholder.
Make ElunaInstanceAI::Load re-initialize if data failed to load.
Also improve error messages by not including the raw data.
Improve storage format of upvalues
Instead of storing the upvalues by name, store by index. A wrapper is
still used in case the upvalue is nil, to prevent holes in the upvalues table.
A special field in the upvalues table, "E", is used to store the index of
the _ENV reference (if there was one). A reference to the current globals
table is set as the upvalue upon decoding.
Remove wrapping from upvalue storing, instead save amount of upvalues
It's now possible to have complete flexibility with cancelling bindings,
i.e. it's possible to re-implement all of the other ways to cancel bindings
using the callback.
This sytem has been designed with complete safety in mind:
- It is safe to call the callback more than once (does nothing after the
first call).
- The callback does not expire, and can be stored for as long as
necessary.
- If a callback is returned, there is no other way to cancel the binding
(again, due to safety).
Because GUIDs are, in fact, not globally unique on mangos, the actual
unique identifier is the GUID/instance ID pair of the Creature.
On Trinity Creatures in instances are assigned new GUIDs. This means that
the instance ID part is redundant but must be used anyway for consistency.
It was previously unsafe to reload Eluna because it had to be reloaded
with the mutex unlocked. This is because the mutex's lifespan was shorter
than the Eluna instance it was guarding.
This allows handlers that return values to see the returned values of previous
handlers. E.g. if Handler A modifies the value from 10 to 20, Handler B
will receive 20 as an argument instead of the original 10.
* Attempt fix VS 2010 pow error
* Tweak the doc parser a little to work with variable arugments ... and show
enum names and similar.
* Change push and check to work with more types like size and time (fix osx)
* Add errors to using not implemented operations on eluna objects
* Change some SQL comments to work with documentation better
* Add functions to create uint64 and int64 values (may need rename)
* Change doc generation documentation a little to make it clearer how proto is used
