llVarName(string str)

I have a suggestion for an LSL function.

llVarName(string str)

Whereby you can pass in a string and it will return a variable's NAME. LSL would use the returned name to either create a variable by that name (if declaring a var), or call the var if it exists (else returns an error if the var was called but it doesn't exist).

What this allows is dynamic variable calls. For example, say I have a game that has 3 players. I keep scores and player keys for each player. I name my vars with a numeric "tag" for ease of scripting (assuming llVarName() existed).

// player keys
key g_sPlayer1;
key g_sPlayer2;
key g_sPlayer3;
// player scores
integer g_nPlayer1;
integer g_nPlayer2;
integer g_nPlayer3;
// player count
integer g_nTotalPlayers;

// Now I can easily make functions/routines such as:
AddPlayer(integer i, key pid)
{
  llVarName("g_nPlayer"+(string)i) = pid; // if "i" was "2" then this statement is evaluated as: "g_nPlayer2 = pid;"
}

default
{
  touch_start(integer num)
  {
    if (g_nTotalPlayers < 3)
    {
      g_nTotalPlayers++;
      AddPlayer(g_nTotalPlayers, llDetectedKey(0));
    }
  }
}

Would also be quite handy for creating multiple dynamically created local vars:

integer i = 0;
while (i >= 0 && i < 10)
{
  i++;
  string llVarName("NewStringNumber"+(string)i) = "Yadda Yadda";
}
// gives you 10 string vars "NewStringNumber1" to "NewStringNumber10"

Anywho, my twisted idea for today!
--Boso

Edit:
Should return a null string "" if the var doesn't exist

if (llVarName("MyVar"+(string)i+"hrm") != "")
{
  // var exists
}
else
{
  // var doesn't exist
}

I know that was a bit of a trivial example you posted, but dont lists provide somewhat the same kind of functionality?
==Chris

What Bos is asking for is a dictionary class.

Which, as Christopher has indicated, can be fairly easily simulated by using a list with a stride of 2, or paired lists.

Sounds like what you want is pointers or symbolic references.

pointers would be very nice.

Strife, indeed. And I think you make the Linden Sysadmins quake in fear mentioning them. Talk about opening a huge hole for people to crash sims...

I dont even understand what boso is asking for, but I would sure love to have the standard eval(string code) function call.

they wouldn't be real pointers in the sense of the word. They would be restricted to the memory space of the script. But it would be cool to use a pointer to reverse engineer the scripting virtual machine. Oh i can't wait . I think it would be fun to make a polymorphic script that could change it's own code by manipulating the copy of the code that is stored in the 16k memory space.

Only if the implementation was bulletproof. I once hacked a stack machine by finding one of the stack manipulation ops that wasn't protected properly, and using very large offsets to access data outside the virtual machine.



And this is a big part of the reason it won't be implemented: self-modifying code is a bitch to debug, and even harder to protect the virtual machine against.

Gee, and the fact that self-modifying code that can replicate is ... oh, a virus? LOL

I think Strife was being a lil' sarcastic

Okay, I think we should stop hijacking this thread and let it die.

-Hiro Pendragon