Little endian / Big endian

[Moru]

I'm having troubles with converting unsigned longs from big endian to little endian format. Is there any bitwise shift command in GL-Basic? I'm using this code now but I can't convert numbers that end with a byte that starts with a 1 (binary).

Code (glbasic) Select Expand

x = MSB_long(0xC8000000)
// x is supposed to be 200 after this call but the result you get is -128

FUNCTION MSB_long: number
LOCAL x1,x2,x3,x4

// Isolate the separate bytes
x1 = bAND(number, 0xff)
x2 = bAND(number, 0xff00) / 0x100
x3 = bAND(number, 0xff0000) / 0x10000
x4 = bAND(number, 0xff000000) / 0x1000000

// Add back the bytes in opposite order
number = 0
number = number + x1 * 0x1000000
number = number + x2 * 0x10000
number = number + x3 * 0x100
number = number + x4
RETURN number
ENDFUNCTIONwhat I want to do is just shift the bytes one at a time to the right instead of dividing by 0x1000000. If I do that he will think I'm dividing a negative value because of the first bit being set. Does anyone know of a better way to do this if there is no way of shifting bits?

[Thunor]

Well this should work but it doesn't

Code (glbasic) Select Expand

FUNCTION MSB_long: number
    LOCAL x1,x2,x3,x4

    // Isolate the separate bytes
    x1 = bAND(number, 0xff)
    x2 = bAND(number / 0x100, 0xff)
    x3 = bAND(number / 0x10000, 0xff)
    x4 = bAND(number / 0x1000000, 0xff)

PRINT x1 + " " + x2 + " " + x3 + " " + x4 + " ", 100, 60

    // Add back the bytes in opposite order
    number = 0
    bOR(number, x1 * 0x1000000)
    bOR(number, x2 * 0x10000)
    bOR(number, x3 * 0x100)
    bOR(number, x4)

PRINT number, 100, 80

    RETURN number
ENDFUNCTIONIt prints :-
0 0 0 200
0

And this prints the same :-

Code (glbasic) Select Expand

FUNCTION MSB_long: number
    LOCAL x1,x2,x3,x4

    // Isolate the separate bytes
    x1 = bAND(number, 0xff) * 0x1000000
    x2 = bAND(number / 0x100, 0xff) * 0x10000
    x3 = bAND(number / 0x10000, 0xff) * 0x100
    x4 = bAND(number / 0x1000000, 0xff)

PRINT x1 + " " + x2 + " " + x3 + " " + x4 + " ", 100, 60

    // Add back the bytes in opposite order
    number = 0
    bOR(number, x1)
    bOR(number, x2)
    bOR(number, x3)
    bOR(number, x4)

PRINT number, 100, 80

    RETURN number
ENDFUNCTIONGotta go to bed now...

[Kitty Hello]

OK, I've fixed this. It had to do with the most significant bit.

However: You're converting a number that's too big to be stored in a 32 bit float. Thus, when you port this program to PocketPC/GP2X, it _will not work_!
If you really must handle this big integers, please go for a wrapper using INLINE.

[bigsofty]

How difficult is it to add your own INLINE types Gernot? Also what should we look out for with regards to comparability with rest of the GLB Functions/System? Could you provide a simple example please?

[Kitty Hello]

I would use GLB types, but perform some INLINE action to convert/work with e.g. 32 bit integral types.
The whole problem usually comes from porting something from a different language to GLBasic. I never had problems of this kind before and I can't think of one out of my head now.
Why would one want to flip the endianess of a integer? Why can't you just store it correctly in the first place?
MAybe I'm missing some point here...

[Moru]

To be as general as possible because the format is written for non-intel systems. You can select big or little endian but default is the non-intel version so I need to be able to convert it.

[Kitty Hello]

Oh dear! GLBasic does the same, but goes the little endian (x86) way for READLONG and such.
I suggest writing a wrapper that uses readfile and reads single bytes. Then make a number from these.

[Moru]

ah, good idea, didn't think that far. Thanks!