[dtolnay]

I interpreted the author's characterization to be about something like:

  1  if (argc <= 2)
  2    puts("A");
  3  puts("B");
  4  if (argc <= 2)
  5    unreachable();
  6  else
  7    return puts("C");
  8  return puts("D");

in which not just lines 4-6,8 go away (as you said) but also lines 1-2.

It makes sense to me but I can see why the author would characterize this situation as "license to use an unreachable annotation on one code path to justify removing an entirely different code path that is not marked unreachable". In a different world one might expect A to be printed "before the UB happens".

[masklinn]

On the other hand, that has been the behaviour of optimising compilers in the face of UBs for years at this point, decades maybe. The linux kernel was hit by a deref' constraint propagation back in 2009 or so.

This is a behaviour I would absolutely expect from the construct, I would even qualify it as "the point".

[rrobukef]

I find this especially surprising because line 2 may be exactly the reason why line 5 is unreachable. E.g. if puts("A") contractually throws an exception you cannot just remove it.

What am I missing in this example?

[zefix]

C does not have exceptions...

[rrobukef]

But it has long_jmp, right?