Second draft: Sum Type by Struct

Richard (Rikki) Andrew Cattermole richard at cattermole.co.nz
Sat Sep 12 18:36:35 UTC 2026


On 12/09/2026 10:42 PM, Per Nordlöw wrote:
> On Friday, 11 September 2026 at 22:12:16 UTC, Richard (Rikki) Andrew 
> Cattermole wrote:
>> On 12/09/2026 10:07 AM, Per Nordlöw wrote:
>>> On Thursday, 10 September 2026 at 17:56:21 UTC, Richard (Rikki) 
>>> Andrew Cattermole wrote:
>>>> No.
>>>
>>> Can you please give a motive to why you don't want to support this? 
>>> And I don't understand why you are pasting the code failure. The code 
>>> works for me.
>> Bit packing requires assumptions.
> 
> And how is that blocker? The optimization could kick only when the 
> alignment of the pointed to types give room to tag bits which is often 
> the case for struct types.

That would require reordering the fields.

The tag goes first.

And quite frankly nobody is implementing this. Its not how struct & 
union field layouting works.

>> The bottom bits are used, it is not a modulas of a the type size.
> 
> For what values of `Types` passed to my `GCWordVariant` are you 
> referring to here?

int, int*


More information about the dip.development mailing list