From: already5chosen@yahoo.com   
      
   On Tue, 25 Nov 2025 11:38:32 +0000   
   bart  wrote:   
      
   >   
   > No, apart from the usual set of 8/16/32/64 bits. I've done 128 bits,   
   > and played with 1/2/4 bits, but my view is that above this range,   
   > using exact bit-sizes is the wrong way to go.   
   >   
      
   Either that or manifestation of your NIH syndrome.   
   Which explanation do you consider more likely?   
      
   > While for odd sizes up to 64 bits, bitfields are more apt than   
   > employing the type system.   
   >   
      
   int sign_extend12(unsigned x)   
   {   
     return (_BitInt(12))x;   
   }   
      
   Nice, is not it?   
   Doing the same with bit fields is possible, but less obvious and less   
   convenient. Also it potentially can play havoc with compiler that took   
   strict aliasing rules more seriously than they deserve.   
      
   int sign_extend12(unsigned x)   
   {   
     struct bar {   
       signed a: 12;   
     };   
     return ((struct bar*)&x)->a;;   
   }   
      
   Doing the same with shifts is almost as convenient as with _BitInt and   
   it works great on all popular compilers, but according to wording of C   
   Standard it is Undefined Behavior.   
      
   int sign_extend12(unsigned x)   
   {   
     return (int32_t)((uint32_t)x << 20) >> 20;   
   }   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)