[fpc-devel] A bit of background on compilers exploiting signed overflow

[Jonas Maebe]

wkitty42 at windstream.net wrote:
> i thought this was a very interesting discussion...
>
> https://gist.github.com/rygorous/e0f055bfb74e3d5f0af20690759de5a7
>
> thanks to john carmack (@ID_AA_Carmack) for retweeting this from Fabian
> Giesen ‏(@rygorous)...

The point is kind of irrelevant in case of FPC, because overflows 
(signed or unsigned) are not undefined here. Every type is guaranteed to 
exhibit the behaviour defined by 2's complement for its particular byte 
size.


Jonas