The combination of variance and implied bounds for nested references opens a hole in the current type system:

staticUNIT:&'static&'static() =&&();fnfoo<'a,'b,T>(_:&'a&'b(),v:&'bT) ->&'aT{ v}fnbad<'a,T>(x:&'aT) ->&'staticT{let f:fn(&'static&'a(),&'aT) ->&'staticT = foo;f(UNIT, x)}

This hole has been fixed in#129021 for non-higher-ranked function pointers. The underlying issue still persists.

staticUNIT:&'static&'static() =&&();fnfoo<'a,'b,T>(_:&'a&'b(),v:&'bT, _:&()) ->&'aT{ v}fnbad<'a,T>(x:&'aT) ->&'staticT{let f:fn(_,&'aT,&()) ->&'staticT = foo;f(UNIT, x,&())}fnmain(){}

Update from@pnkfelix :

While the test as written above is rejected by Rust today (with the error message for line 6 saying "in type&'static &'a (), reference has a longer lifetime than the data it references"), that is just an artifact of the original source code (with its explicit type signature) running up againstone new WF-check.

The fundamental issue persists, since one can today write instead:

staticUNIT:&'static&'static() =&&();fnfoo<'a,'b,T>(_:&'a&'b(),v:&'bT) ->&'aT{ v}fnbad<'a,T>(x:&'aT) ->&'staticT{let f:fn(_,&'aT) ->&'staticT = foo;f(UNIT, x)}

(and this way, still get the bad behavingfn bad, by just side-stepping one of the explicit type declarations.)

Update from@lcnr :

While the test as written above is rejected by Rust today, that is just an artifact of the original source code (with its lack of higher ranked regions) running up againstone new WF-check ✨

The fundamental issue persists, since one can today write instead:

staticUNIT:&'static&'static() =&&();fnfoo<'a,'b,T>(_:&'a&'b(),v:&'bT, _:&()) ->&'aT{ v}fnbad<'a,T>(x:&'aT) ->&'staticT{let f:fn(_,&'aT,&()) ->&'staticT = foo;//                  ^ note the additional higher-ranked region heref(UNIT, x,&())}

All the non-higher ranked variants of this issue have been fixed by#129021.