Binary OR width increase

ID:105817
 
Dec 16 2010, 9:28 pm (Edited on Dec 19 2010, 7:37 pm)
Applies to:DM Language
Status: Open
Issue hasn't been assigned a status value.
DM's binary OR width is currently 16 bits which only allows proper ORing of values from 0 to 65535. Please increase the upper bound to 232-1.

This works in the meantime.
proc{
    _32bit_or(a,b){
        var{
            a1=round(a/(2**16));
            a2=a%(2**16);
            b1=round(b/(2**16));
            b2=b%(2**16);
        }
        .=((a1|b1) * (2**16)) + (a2|b2);
    }
}
Dec 17 2010, 5:43 am
Looks like all binary operations are done with 16-bit integers. Luckily a 32-bit binary operation can easily be split into two 16-bit operations.
Apr 7 2011, 9:11 am
Increasing to 32-bit precision is impossible because the internal number format is floating point, which only allows for 24 bits of precision. I'm not averse to increasing precision to 24-bit though it could impact existing games.