Complex If statement processing

[SKKmods]

With a bunch of complex AND logical tests in a single If statement, does the papyrus script engine bail on the first test fail, or process the whole line, so better to nest them ?

 

Neat for code maintenance:

 

If (Simple logical test A) && (Simple logical test B) && (Simple logical test C) && (Complex array lookup logical test D)

;do stuff

EndIf

 

Better for performance ?

 

If (Simple logical test A)

If (Simple logical test B)

if (Simple logical test C)

If (Complex array lookup logical test D)

;do stuff

Endif

Endif

Endif

Endif

 

 

Or halfway split simple and complex ?

 

If (Simple logical test A) && (Simple logical test B) && (Simple logical test C)

If (Complex array lookup logical test D)

;do stuff

Endif

Endif

 

Because performance is a virtue.

[Reneer]

I'm pretty sure the compiler takes care of that - if the first condition fails the others are not evaluated. There is a Papyrus performance page on the CK Wiki: https://www.creationkit.com/fallout4/index.php?title=Performance_(Papyrus)#Fail_fast.2C_fail_early Edited June 1, 2018 by Reneer

[pra]

I'm pretty sure the compiler takes care of that - if the first condition fails the others are not evaluated. There is a Papyrus performance page on the CK Wiki: https://www.creationkit.com/fallout4/index.php?title=Performance_(Papyrus)#Fail_fast.2C_fail_early

It's not the compiler, but the interpreter/vm/whatever you want to call the part which actually runs the .pex files </nitpick>

 

I think it can be safely assumed that most programming languages are able to do this sort of short-circuiting, but thank you for confirming it nonetheless.

 

That said, I sometimes make several ifs with return statements instead of one single one for the sake of readability. But also if I need a variable for one of the checks:

function doStuff(ObjectReference someRef)
    
    if(!someRef)
        return
    endif
    
    ObjectReference player = Game.GetPlayer()
    if(someFunction(player, someRef) = 0) 
        return
    endif
    
    ...


endfunction

[SKKmods]

Thanks Foxy, with the lack of proper sdk specs and docs, optimisations are rather opaque. I shall find some time to setup a benchmark for this, a useful investment in a (probably) sunset platform ...

 

@pra I try and avoid using return as that just leads to me creating complex nested chains of recursive functions that stack and crash.

[pra]

 

@pra I try and avoid using return as that just leads to me creating complex nested chains of recursive functions that stack and crash.

Wat? How?

 

I always try to return as soon as possible. If I know that the function can't do anything sensible anymore, I make it return right away. The alternative would be to create complex nested if/else constructs, and having an extra variable for the return value, if any. Which I find incredibly ugly and hate Pascal (fo4edit scripting language) for trying to enforce it.

[SKKmods]

I'm talking papyrus. not using return forces me to chunk my functions and use events rather than create thousand line monster functions. its probably philosophical ...

[Evangela]

I personally find the halfway one to get real confusing when it's more than 2 && conditions, with each check being a condition in itself(A > B && A < D && C > D etc..). If chains I can see a "staircase" clearly and not get lost. If there were switches though, I'd probably never need a long if chain. Maybe in the next game..

Edited June 1, 2018 by Rasikko

[Reneer]

 

I'm pretty sure the compiler takes care of that - if the first condition fails the others are not evaluated. There is a Papyrus performance page on the CK Wiki: https://www.creationkit.com/fallout4/index.php?title=Performance_(Papyrus)#Fail_fast.2C_fail_early

It's not the compiler, but the interpreter/vm/whatever you want to call the part which actually runs the .pex files </nitpick>

 

True enough, since the compiler can't exactly foresee what the variables are going to be at runtime. :tongue:

[FatsackTony1]

Every language I know short circuits as soon as the first false is detected when you're comparating a bunch of and statements.

[pra]

 

 

I'm pretty sure the compiler takes care of that - if the first condition fails the others are not evaluated. There is a Papyrus performance page on the CK Wiki: https://www.creationkit.com/fallout4/index.php?title=Performance_(Papyrus)#Fail_fast.2C_fail_early

It's not the compiler, but the interpreter/vm/whatever you want to call the part which actually runs the .pex files </nitpick>

 

True enough, since the compiler can't exactly foresee what the variables are going to be at runtime. :tongue:

 

It could do compile-time optimisation, though. For example, if you literally write "if (foobar && true)" it could optimize the second part away. A more advanced compiler could also understand something like "bool foo = true; if(foo)".

 

But I doubt this bethesda compiler could do anything like this.