SM Dynamic Array Float Math Issue

Malachi · **Author**

So I had this crazy problem yesterday that I just figured out, and wanted to know if this is an issue or if I just missed something.

Works:

Code:

new Float:A[SIZE];
...
A[Index] += 25.0;

Doesn't work:

Code:

new Float:A[SIZE];
    myArray = CreateArray(1);
    PushArrayCell(myArray, 25.0);
...
A[Index] += GetArrayCell(myArray, Index);

Also doesn't work:

Code:

new Float:A[SIZE];
    myArray = CreateArray(1);
    PushArrayCell(myArray, 25.0);
...
A[Index] = GetArrayCell(myArray, Index) + A[Index];

Also also doesn't work:

Code:

new Float:A[SIZE];
    myArray = CreateArray(1);
    PushArrayCell(myArray, 25.0);
...
A[Index] = float(GetArrayCell(myArray, Index)) + A[Index];

But this works:

Code:

new Float:A[SIZE];
    myArray = CreateArray(1);
    PushArrayCell(myArray, 25.0);
...
A[Index] = FloatAdd(GetArrayCell(myArray, Index), A[Index]);

So I've never seen that before. Is this a problem w/SM? Or am I just missing something?

And is this the proper way to do this?

Thanks!

Malachi · 05-27-2014 , 15:42 Re: SM Dynamic Array Float Math Issue

Almost forgot:
SourceMod Version Information:
SourceMod Version: 1.5.3
SourcePawn Engine: SourcePawn 1.1, jit-x86 (build 1.5.3)
SourcePawn API: v1 = 4, v2 = 4
Compiled on: Mar 21 2014 07:50:09
Build ID: 4026:53569c20fc33
http://www.sourcemod.net/

Metamod:Source version 1.10.0
Build ID: 860:a58a1912f602
Loaded As: Valve Server Plugin
Compiled on: Aug 25 2013
Plugin interface version: 15:14
SourceHook version: 5:5
http://www.metamodsource.net/

TF2 linux server.

Powerlord · *Last edited by Powerlord; 05-27-2014 at 16:24.*

When you pull a Float value back out of an adt_array, you must use the Float tag:

PHP Code:


			
Float:GetArrayCell(myArray, Index)

If you don't, it thinks you're pulling back an integer and will then try to convert said integer value to a Float value using float() (int + float always promotes the int to a float first and returns a float), which doesn't resemble the original value at all because it already was a Float but SourceMod didn't know it.

So, likely what you want is this:

PHP Code:


			
new Float:A[SIZE];
    myArray = CreateArray(1);
    PushArrayCell(myArray, 25.0);
...
A[Index] += Float:GetArrayCell(myArray, Index);

Bacardi · *Last edited by Bacardi; 05-27-2014 at 16:12.*

nvm

Powerlord · *Last edited by Powerlord; 05-27-2014 at 17:14.*

As for why float() doesn't work deals with how the number formats work. Both integer and Float in pawn are 32-bits long.

integer (the normal cell value type) is:
1 bit for sign
31 bits for value

Float is:
1 bit for sign
8 bits for exponent (power + 127)
23 bits for the Mantissa/significant... or the remaining value

For now, lets assume we're dealing with the number 1 (or 1.0). Here's how it looks in binary

0000 0000 0000 0000 0000 0000 0000 0001

It works a bit differently for floating point...

Since we already know the binary representation, we take that, convert it to scientific notation (1.0 * 2^0) and add 127 to the power (0 + 127 = 127)... so....
0 01111111 00000000000000000000000
(remember the "1." is dropped from the mantissa because it's part of the power)

If you treat the representation of 1.0 as a integer, you end up with 1065353216 and float(1065353216) just gives you 1065353216.0 (1.111111 * 2^29 or 0 10011100 11111100000000000000000) instead of 1.0.

Malachi · 05-27-2014 , 17:23 Re: SM Dynamic Array Float Math Issue

Quote:

Originally Posted by Powerlord

When you pull a Float value back out of an adt_array, you must use the Float tag:

PHP Code:


			
Float:GetArrayCell(myArray, Index)

Ah crap, I actually looked for that (I knew about the _: tag) - but didnt' find it.

Ok, I'm going to try that (and hopefully feel like things make sense again).

Thanks!
(and a very fast reply too)

Malachi · 05-27-2014 , 17:39 Re: SM Dynamic Array Float Math Issue

Confirmed that this appears to be working.

Thanks again!