Var tracking :D?

ID:132971
 
Jul 11 2009, 8:12 pm (Edited on Jul 12 2009, 7:04 am)
 
Var_Accessed(var_name,atom/Src) // Src is the owner of the var, if available (would be null for global vars and stuff)
 world<<"[var name] accessed!"
 switch(var_name)
  if("h_pee")
   if(ismob(Src))Src:UpdateHP()

mob/var/h_pee=1337
mob/verb/ChangeHP(a as num)
 usr.h_pee=a // the var has been accessed! Var_Accessed() would be called.
mob/proc/UpdateHP()
 winset(w.e)


Gets called whenever a var is accessed. This would simplify certain things so much. Instead of tracking every single var change by ourselves, just override that prawk.

(Turn vars into procs ftw? Is it possible?)
Jul 11 2009, 8:26 pm
I suppose you could use a datum to handle your variables. Just have a "set" proc to set the value to something, or an "add" proc to add/subtract a number from it(if it's a number).
Jul 11 2009, 8:45 pm
I'm not sure what you're asking, but if it's tracking what user has what variables.

mob
    verb
        UserVariables()
            for(var/A in src.vars)
                src << "[A] = [src.vars[A]]"


That can be adapted for all atoms, as I'm fairly sure every atom has the vars list.
Jul 11 2009, 9:02 pm
In response to Tiberath
I think he's asking for a proc to be called when a variable's value is changed.
Jul 11 2009, 9:18 pm (Edited on Jul 11 2009, 9:28 pm)
In response to Kaiochao
Kaiochao wrote:
I think he's asking for a proc to be called when a variable's value is changed.

That's easy enough to achieve without the use of datums.

mob
    proc
        update_variable(variable, value)
            if(variable in src.vars) 
                src.vars[variable] = value
                src << "Your [variable] has been changed to [value]!"
                // Perform additional code here.
            else return FALSE
            
mob
    var
        a = 0
        
    verb
        test()
            var/name = input("Please select your name", "Name") as null|text
            if(name)
                src.update_variable("name", name)
                world << src.name
                
        test2()
            // Same as src.a += 36
            src.update_variable("a", src.a + 36)


Of course, it gets tedious if you want to perform different actions per variable, but still doable. It just means you can't set the variable as you normally would. (if you want it to be for all atoms, just change the mob/proc to atom/proc and away you go.)
Jul 11 2009, 9:34 pm
In response to Tiberath
That's not exactly what I meant :/

Whenever a var is accessed the proc is called automatically with the info. Of course we can do this manually, but we could just override the Var_Accessed() proc, and not worry about calling the specific proc for every single var we change. I have trouble expressing myself sorry D:
Jul 12 2009, 7:03 am
seconded. maybe a proc for datums? something like _var_changed(variable, oldvalue, initialvalue)?

helps great in converting old projects!
Jul 12 2009, 12:56 pm
This sounds like the equivalent of properties in some other languages. A syntax similar to Python's would be better (although probably harder to implement) than having one global proc that is called every time any variable is accessed.
Jul 14 2009, 4:44 pm
I don't think this would be too hard to implement...
Just have a proc call whenever any of the assignment operators (= += -= *= /= &= |= ^= <<= >>=) are used during run time.
Jul 14 2009, 5:36 pm
In response to Naokohiro
Naokohiro wrote:
I don't think this would be too hard to implement...
Just have a proc call whenever any of the assignment operators (= += -= *= /= &= |= ^= <<= >>=) are used during run time.

This would be even easier to just program yourself, taking an OOP approach.

mob/MyPlayer
   proc/LoseHP(N)
      HP-=N
      DeathCheck()


Then at any time while you're programming if they lose health just use LoseHP(Damage) instead of HP-=Damage.

It is a very good practice to have anyway while programming. If you ever work in a language like Java most of the tutorials you read encourage keeping variables private (only able to be modified from inside the module you're working on) and changing them externally through public functions (procs).

It isn't that I don't think this is a good idea but honestly you can do it yourself, and it isn't hard.
Jul 14 2009, 5:38 pm
In response to AJX
That's not an object-oriented approach. That's a functional approach. An object-oriented approach would be creating your own class (i.e., a datum) to handle the variable information.
Jul 14 2009, 5:59 pm
In response to Popisfizzy
Popisfizzy wrote:
That's not an object-oriented approach. That's a functional approach. An object-oriented approach would be creating your own class (i.e., a datum) to handle the variable information.

Duly noted. But the point still stands. :)
Jul 14 2009, 6:11 pm
In response to AJX
I know that. It's quite obvious how you can program it to do this, but it's just not pretty, nor will it work for existing projects that haven't already used this method throughout the code.
Jul 14 2009, 6:15 pm
In response to Naokohiro
Naokohiro wrote:
I know that. It's quite obvious how you can program it to do this, but it's just not pretty, nor will it work for existing projects that haven't already used this method throughout the code.

To be honest I don't think it would be difficult in the slightest to apply this sort of system to an existing project. A simple find/replace of HP+= / HP-= to HPUp( / HPDown( and throwing a ) on after each one would only take a few seconds in a small project, and minutes in a large project.

And I also don't believe it isn't pretty either. As a personal preference I MUCH prefer using a single proc for handling damaging a player which automatically calls deathcheck() than doing HP-=5 , DeathCheck() in every code where someone could take damage.


I just personally feel that the developer's time is better spent on things that can't be easily managed by ourselves. Sure, convenience is nice but not in such a simple situation and not at the cost of their valuable time.
Jul 14 2009, 6:23 pm
In response to AJX
I guess what I'm trying to say is that I'm not looking for a workaround. Unless this is the real thing, then I don't care about it, nor will I consider how I could use var tracking anyways. See ya.