[gNewSense-users] Scummvm games: software or data?

[Mark H. Weaver]

It has been claimed that the scummvm games are merely data, and thus it is 
okay to keep them in gNewSense even without source, the preferred form for 
editing.  That is a reasonable claim, but upon further investigation I 
believe it is false.

Take a look at struct VirtualMachineState, in scummvm-0.8.0/scumm/script.h:

  struct VirtualMachineState {
        uint32 cutScenePtr[5];
        byte cutSceneScript[5];
        int16 cutSceneData[5];
        int16 cutSceneScriptIndex;
        byte cutSceneStackPointer;
        ScriptSlot slot[NUM_SCRIPT_SLOT];
        int32 localvar[NUM_SCRIPT_SLOT][26];

        NestedScript nest[15];
        byte numNestedScripts;
  };

So, we have local variables and nested scripts.  There's a debugger too.
Line 66 of scummvm-0.8.0/doc/05_04.tex tells how to enter the debugger in a 
section of special keys specifically for Flight of the Amazon Queen.

Take a look at scummvm-0.8.0/scumm/script.cpp, which has the following 
function:

  /* Start executing script 'script' with the given parameters */
  void ScummEngine::runScript(int script, bool freezeResistant,
     bool recursive, int *lvarptr, int cycle) {
  [...]

So we have recursion, local variables, and parameters.

In the same directory, look at the various script_v*.cpp files, for example 
script_v6.cpp, which contains a list of opcodes.  Here are some selected 
opcodes, to give you an idea: o6_pushByte, o6_pushWord, o6_pushByteVar, ..., 
o6_dup, o6_not, o6_eq, o6_neq, o6_gt, o6_lt, o6_le, o6_ge, o6_add, 
o6_sub, ..., o6_pop, o6_byteArrayIndexedWrite, o6_if, o6_ifNot, 
o6_startScript, ..., o6_jump, ...

I therefore claim that scummvm games are not merely data, but software.

     Mark