Re: Hex string literals (was Re: C23 thoughts and opinions)

[...] I noticed this strange discrepancy between
between my hex floats and C's.  In C, this value:
>
    0x12.34p10
>
has the decimal value 18640.0; why was that?  It turns that this is:
>
   18.203125 x 2 ** 10
>
The main part is interpreted as actual hex;  the exponent is in
decimal, and exponent scaling is in binary bits.  So THREE diferent
bases are involved!
>
In my scheme (which as I said worked for bases from 2 to 16), the
0x12.34p10 value would mean this:
>
   18.203125 x 16 ** 16        ~= 3.58e20
>
The same base is used for all parts, including the exponent, and the
digits that are to be shifted.  After all in decimal, 12.34e10 would
mean:
>
   12.34 x 10 ** 10            = 123400000000.0
>
In my language, a base 5 version 5x12.34e10  (not valid below base 5)
would mean this (I think, as I no longer have the compiler):
>
   7.76 x 5 ** 5               = 24250.0
>
This has a consistency lacking in C's hex floats.