- Removed the function number parameter from all kernel functions, as it's no longer needed, and removed the FAKE_FUNCT_NR hack
- Removed kUnknown() and kStub()
- Dummy/unknown kernel functions are no longer invoked, and a warning is shown instead, with the paremeters passed to them
Note: there is an evil hack used for debugging scripts in invoke_selector(), which probably no longer works now
svn-id: r44461
It did not undo all effects of a script being deleted (specifically its
superclasses remain unlocked), causing an inconsistent
state. Also removed EcoQuest CD hack that worked around a specific
instance of this problem.
svn-id: r44449
- rename segGet and getSegment to getScriptSegment; the two can be
distinguished by the parameter count.
- rename type SCRIPT_GET to ScriptLoadType to conform with our code
formatting conventions
- rename get_class_address to getClassAddress
- some cleanup
svn-id: r43981
- Added detection for the EcoQuest 1 demo in the fallback detector
- Partially reverted my previous "fix" for EcoQuest 1 CD, and turned it into a script-specific hack for that game, for now
- Added handling of kDisposeScript calls made with 2 parameters, e.g. in KQ5CD and others (still untested, haven't found a test case)
svn-id: r43887
- Replaced some EngineState parameters
- The SCI version is now obtained from the resource manager or the segment manager, thereby simplifying several functions
- Plugged 2 leaks in the fallback detector
- Renamed the segment manager and resource manager to "segmentManager" and "resourceManager" in all places, for consistency
svn-id: r43722
- Moved the class table in the Segment manager - it's the only class using it directly
- Removed the sci11 flag from save games (we already know this, we don't need to store it)
- Moved script_get_segment() and get_class_address() inside the segment manager class
- Removed the script_locate_by_segment wrapper
- Simplified script_lookup_export() a lot by removing some paranoia checks
- Added some WIP code for automatically determining the game id in the fallback detector (still not working)
- General cleanup
svn-id: r43458
Storing them as reg_t* could cause the pointers to become invalid
when they pointed into CloneTable since CloneTable can be re-allocated.
svn-id: r41220
- Rewrote the sci_opcodes enum so that it's easier to read
- Made the engine error out if data is sent to an invalid selector (which is a fatal condition)
svn-id: r41069
- Turned all SCIkwarn and SCIkdebug functions to ScummVM's debugC function
- Placed some debug code in appropriate defines: DEBUG_PARSER, DEBUG_AVOIDPATH and DEBUG_SOUND
- Removed the "debuglog" command and the "script_checkloads_flag" and "sci_debug_flags" variables
svn-id: r41033
