Suggestion for JESP standard

03-30-2007, 01:15 AM

Since I didn't know where else to post this, I decided here was as good a place as any. With the dawn of precompilers, why is it that for configuration sections of spells we still use functions. I would think that it would be more logical, and it would greatly increase run time by decreasing the amount of function calls, to use globals. For example:

JASS:

//################## CONFIGURATION SECTION ###################

function Eviscerate_Configuration takes nothing returns nothing
    set Eviscerate_Damage[1] = 25
    set Eviscerate_Damage[2] = 35
    set Eviscerate_Damage[3] = 50
    set Eviscerate_DamageCap[1] = 50
    set Eviscerate_DamageCap[2] = 65
    set Eviscerate_DamageCap[3] = 90
endfunction

globals
    integer Eviscerate_SpellId = 'A000' // This is the raw code of the spell
    integer Eviscerate_BuffId = 'B000' // This is the raw code of the buff
    real Eviscerate_Range = 220 // This is the range in which the damage will be dealt


//################## END OF CONFIGURATION SECTION ###################

    //For data which depends on the level of the ability, make an array
    //and set it in a "Configuration Function" which is run at map init

    real array Eviscerate_Damage 
    real array Eviscerate_DamageCap
endglobals

//########################### SPELL SECTION ##########################
...

JASS:

...

function InitTrig_Eviscerate takes nothing returns nothing
    set gg_trg_Eviscerate = CreateTrigger(  )
    call Eviscerate_Configuration()
    call TriggerRegisterAnyUnitEventBJ( gg_trg_Eviscerate, EVENT_PLAYER_UNIT_SPELL_EFFECT )
    call TriggerAddCondition( gg_trg_Eviscerate, Condition( function Eviscerate_Conditions ) )
    call TriggerAddAction( gg_trg_Eviscerate, function Eviscerate_Actions )
endfunction

The one thing I am NOT sure about, however, is whether it is faster to reference an index of an array than to pump a value through a function. One pro of this change is that spells with non-linear data can have their data put in level by level, so that makes configuring those types of spells easier. One con though, is that it is a lot more confusing to read, and possibly more confusing to write. And, of course, it would require a precompiler. Hopefully the increase in performance can outweigh the cons.

03-30-2007, 06:24 AM	#2
Captain Griffen	Inliners FTW. Constant functions get inlined by Vex's optimiser, so just run it through that and you can have the best of both world - speed and non-linear data.

03-30-2007, 11:40 AM	#3
Rising_Dusk	Also, you don't need to write the function call as you did -- Just use initializer functions (Since you're using vJass anyways, it appears).

03-30-2007, 11:45 AM	#4
Toink	Is this from AotZ? Zhall's Evis?

03-30-2007, 12:03 PM	#5
Rising_Dusk	Hunh? No. Zhall's Evis was coded long before vJass was released. It's not like Eviscerate is a unique spell name anyway. :P

03-30-2007, 01:15 PM

Vexorian

The new JESP standard is a lot more flexible and is read as (has a config header, text name replace creates a new instance)

0.9.7.2 's changes to scopes(and therefore libraries) threaten to kill the JESP standard since they do this all in a very easy way.

JASS:



//notice how just changing the scope name creates a different instance of the spell template
scope SpellCodeName

   globals
       private constant real delay = 4.5
   endglobals

 //=================================
    private function dostuff takes unit u returns nothing
         call KillUnit(u)
    endfunction


    private function onSpellEffect takes nothing returns nothing
         call PolledWait(delay)
         call SetTableReal("["+SCOPE_PRIVATE+"]", "..." , 3.0 )
         //did you know that above is converted to "[SpellCodeName?__]" automatically?
         call dostuff()
    endfunction


    public function InitTrig takes unit u returns nothing //Becomes InitTrig_SpellCodeName...
     local trigger t=CreateTrigger()
 
        call TriggerAddAction(t, function onSpellEffect)
        call TriggerRegisterAnyUnitEventBJ(t, EVENT_PLAYER_UNIT_SPELL_EFFECT)
        // no need to set to null since it is never destroyed
    endfunction
   

endscope

03-30-2007, 11:21 PM	#7
Szythe	Very enlightening. I had no idea the full extent of how powerful vJass really is. Thanks for the info.