The idea behind  QTBUG-100485 and QTBUG-100486 is:

The idea behind the Q_CONSTINIT and constexpr epics was to mark every static and thread_local variable constexpr, if it can be. Failing that, mark it Q_CONSTINIT const. Failing that, mark it as Q_CONSTINIT. Failing that, mark it as intended dynamic initialization (where we're still missing a standard marker for this).

Then have Clazy or similar do static checks and complain about unmarked statics or thread_locals.

Along the way, try to bring variables into higher categories.

I acknowledge that a Q_CONSTINIT variable may still be expensive to destruct, and therefore construct, because of the required registration of the dtor with the runtime, but we shouldn't let the perfect be the enemy of the good here. Don't try to be clever with PODs and Core Generalized Constant Expressions. These are things for the compiler to check, not for reviewers and other code readers to fuss about. The job of the developer is to express intent, which means exactly one of constexpr, constinit, or Q_DYNINIT(?) should be marking up every static or thread_local variable. No exceptions.

 (quote from mmutz) 

 Providing a clang-tidy modernizer will help to automate marking everything that can be constexpr with constexpr.