This problem looks similar to signed vs. unsigned contradiction: Not a big surprise for those who are familiar with FASM internals, though.

Never thought how fasm does number comparison.., it is signed.
b-a = 10b - 01b = 01b which is unsigned.
All is fine.

serfasm,

If b and a are signed, subtraction overflows and signed result is incorrect. If they are unsigned, b>a.

Fasm should have his own flags register and some interface to access those. Poor Tomasz!

Never thought how fasm does number comparison.., it is signed.
b-a = 10b - 01b = 01b which is unsigned.
All is fine.

signed and unsigned differ only by flags checked.
signed ops use SF and OF, unsigned use only CF.
above/bellow - unsigned, greater/less - signed.


Fasm should have his own flags register and some interface to access those. Poor Tomasz!
fasm its just a link between you and cpu, it cant have anything. flags register is a part of CPU. anyway FASM is already broken in few places, because it use high level encoding. you cant encode eevry instruction with it, and its sad.

sad

erf and lol.

fasm don't have its own flags.
fasm is coded with fasm.
it uses the X86 flags regicter inside your CPU.

the code you compile with fasm does not differ to the code used to compile.

it is all assembler.

i don't see the use of this off topic subject.

hll approach. BTW fasmd.asm compiles in 0.1 seconds as uses no macros, Delphi uses lot-a-lot high level "macro" but no speed consumption(background?).

b = a+1 <- should fasm inform OF=1?

> < are signed or unsigned?

of will be only when yoy cross sign boundary.
for example, try to add something to (byte mode) 127, or substract something from -128.


-128 + 1 = -127, OF = 0
-128 - 1 = 127, OF = 1
127 + 1 = -128, OF = 1
127 - 1 = 127, OF = 0