Re: In precision typing we trust

["Lawrence D’Oliveiro" <ldo@nz.invalid>]

On Sun, 31 Aug 2025 08:51:29 -0000 (UTC), Kevin Chadwick wrote:

> On 31/08/2025 08:30, Lawrence D\x19Oliveiro wrote:
>>
>> What’s missing?
>
> Complete size controls and things like vomatile_full_access and
> proper type protections ...

Want different, incompatible BitVector classes? OK here’s use of a
class factory function that creates such new classes on demand:

    BitVector1 = def_bitvector_class("BitVector1")
    BitVector2 = def_bitvector_class("BitVector2")

First type still works as before:

    v1 = BitVector1(3, 2)
    v2 = BitVector1(5)
    v1, v2, int(v1), int(v2)

    (BitVector1(3, [False, True, False]),
     BitVector1(5, [False, False, False, False, False]),
     2,
     0)

    v1[0:2], v1[2], v1[1:3], v2[2:5]

    (BitVector1(2, [False, True]),
     False,
     BitVector1(2, [True, False]),
     BitVector1(3, [False, False, False]))

The second type works just as well:

    v3 = BitVector2(5)

    v3[:] = 1
    v3, int(v3)

    (BitVector2(5, [True, False, False, False, False]), 1)

but is not directly compatible with the first type:

    v2[:] = v3

    ---------------------------------------------------------------------------
    TypeError                                 Traceback (most recent call last)
    Cell In[5], line 1
    ----> 1 v2[:] = v3

    [rest of traceback omitted]

    TypeError: invalid value for BitVector1 assignment: BitVector2(5, [False, False, False, False, False])

Of course explicit conversion via int does work:

    v2[:] = int(v3)
    v2

    BitVector1(5, [True, False, False, False, False])

Is that some of what you had in mind? Not sure what the rest meant.