Re: netbsd-11 gcc bug

On 5/26/26 4:02 PM, Jason Thorpe wrote:
> 
>> On May 26, 2026, at 7:52 AM, Jörg Sonnenberger <[email protected]> wrote:
>>
>> On 5/23/26 8:47 PM, Manuel Bouyer wrote:
>>> I guess there's still variants of offsetof() around which uses this.
>>
>> offsetof() follows somewhat different rules as it is explicitly an unevaluated context.
> 
> Sure, but some compilers also get upset with the classical definition of offsetof(), and the only way to really fix it is to use the compiler intrinsic.

The rules are different again for C++, which was the original reason for 
switching to the builtin for offsetof. The other reason being much 
better diagnostics for actually illegal uses.

Joerg