[BAILOPAN]

Hi, everyone! Lately I've been studying the SourcePawn compiler. There are a bunch of interesting "features" - some of them are bugs - and, all of them are unsupported and will be removed. But at this specific point in time, they work, and they're amusing.

Today's horrible pawn features are: enums.

Have you ever wanted to make an enum where every index grew by a scale of 73 each time? No? Never? Well, you can:

enum (*= 73) {
    Item1 = 1, // Value is 0
    Item2,     // Value is 73
    Item3,     // Value is 5,329
    Item4      // Value is 389,017
}

Don't forget to initialize that first value to 1, otherwise everything will be 0! Note, you can also specify floats as the multiplier. For example, you can specify "1.5". Obviously this means Item2's value will be "1,069,547,520" and Item3 and Item4 will be 0. Makes sense, right?

You can also place floats inside enums. Did you forget to tag them? Don't worry! It doesn't matter. It's not type-checked.

enum {
    A = 5.5,
    B,
    C
}

Here, the value of "A" is 0x40b00000. "B" is 0x40b00001, and "C" is 0x40b00002. I hope that's clear.

You can also locally scope enums:

public method(a)
{
    if (a) {
        enum MyEnum {
            MyEnum_A,
            MyEnum_B
        }
    }
}

Cool, right? Hold up - what if you do this?

public method(a)
{
    if (a) {
        enum MyEnum {   A, B    }
    } else {
        enum MyEnum {   A, B    }
    }
}

You get:

test.sp(4) : warning 201: redefinition of constant/macro (symbol "A")

[RedSword]

I really don't get this :

Originally Posted by BAILOPAN [...] Note, you can also specify floats as the multiplier. For example, you can specify "1.5". Obviously this means Item2's value will be "1,069,547,520" and Item3 and Item4 will be 0. Makes sense, right?

Does it gives 0 because the 32 last bits resulting of "0011 1111 1100 0000 0000 0000 0000 0000"^2 = 0 ?

Also, I did use PrintToChatAll using 1069547520 * 1.5 and it gave : 1321156608 (wtf... ; %d) or 1604321280.000000.

How is the enum thing not having the same behaviour as the crappy int multiplication ?

Also why do I get something like 1321156608 from that 1.5 multiplication ?!

Otherwise, thanks for sharing.

[berni]

The float multiplier is probably read as normal integer by sourcepawn.

[BAILOPAN]

RedSword: Sorry, I was being facetious there ;) Yeah, it's total nonsense. berni's right. The compiler takes the bits that represent the float and interprets them as an integer.

[Mathias.]

I want the part 2, this is so amusing.

How ever why sourcemod decided to use sourcepawn instead of c++ like amx? It could had compatibility plugin between hl1 and hl2 engine, that would be fun!

[Thraka]

I would so love a different scripting language.

[asherkin]

Originally Posted by Black-Rabbit How ever why sourcemod decided to use sourcepawn instead of c++ like amx? It could had compatibility plugin between hl1 and hl2 engine, that would be fun!

AMX uses Pawn (which is where all these come from), SourcePawn is Pawn improved by the SourceMod team.

[Mathias.]

Oh, someone told me it was c++. Thank to know, but why there are structure and not in sourcepawn? It would be realy interressing.