@opensdh
Beyond the possibly-cosmetic addition of "with automatic storage duration" to p3, I'm wondering whether the "active at Q but not at P" should mention the initialization of local statics somehow. (Those are initialized as control passes through their declaration, too.)

It seems p3 does not intend to cover the block variable with static storage duration since transferring the control from P to Q that is not the valid scope of the variable does not destroy the variable(with static storage duration).

[stmt.dcl#4] is used to phrase the initialization of the block variable with static storage duration while [basic.start.term] covers the destruction of that variable.

Dynamic initialization of a block variable with static storage duration or thread storage duration is performed the first time control passes through its declaration;

In other words, the initialization of the block variable with static storage duration does not occur every time the control passes through its declaration, this is the divergence between the initialization mentioned in p3 and it.

Agreed. p3 should only cover variables with automatic storage duration. Maybe it helps to eliminate the paragraph break between p2 and p3, so that the definition of "active" is directly before its use.

It seems plausible to have /3 destroy only variables with automatic storage duration but initialize all of them, which would replace /4's "first time control passes through" which doesn't really have any definition. However, the rest of that paragraph doesn't seem particularly easy to merge, so maybe it's better to leave it alone and just merge the paragraphs as suggested (since "active" isn't meant to apply to any other kind of variable anyway).