RFR: 8377895: Create sizeof_auto, to reduce narrowing conversions [v7]

[Dean Long]

On Fri, 20 Feb 2026 06:47:24 GMT, Kim Barrett <kbarrett at openjdk.org> wrote:

>> src/hotspot/share/utilities/globalDefinitions.hpp line 179:
>> 
>>> 177:         std::conditional_t<N <= std::numeric_limits<uint32_t>::max(), uint32_t, uint64_t>>>;
>>> 178:   return static_cast<unsigned_auto>(N);
>>> 179: }
>> 
>> This function doesn't really have anything to do with sizeof, does it?  Well, except that it is unsigned.  This could be named something like smallest_type_for_constant, narrow_const, etc.  It is conceivable that in the future we might want something like this for signed values as well.
>
> I think that it can be renamed or otherwise adjusted later if someone comes up
> with other use-cases. But for now, it is provided as nothing more than
> supporting the implementation of sizeof_auto.

OK, so we'll end up with different macros sizeof_auto() and xyz_auto() instead of allowing something more general like auto_size(xyz()) at the call site?

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/29716#discussion_r2835359748