From: gregmmarr@googlemail.com   
      
   On Monday, August 12, 2013 10:30:12 AM UTC-4, SG wrote:   
   > I don't see why unique_ptr should support   
   >   
   >    p.reset(p.get());   
   >   
   > reset is supposed to take a nullptr or pointer to an object that's not   
   > already owned by someone else. So, if you invoke reset with an address   
   > of some object that another (or the same) unique_ptr instance already   
   > owns, you did something wrong.   
      
   Ah, the *not already owned* was probably the part that I wasn't considering.   
   That makes it more obvious why the p.reset(p.release()) was proper but   
   p.reset(p.get()) wasn't.   
      
   Thanks.   
      
   > Can you come with with an example that relies on a self-reset-test but   
   > is not considered broken by others?   
      
   Given the above, probably not.  I guess it was just the fact that something   
   that *seemed* to be allowed by the draft standard (yes Daniel, I realize that   
   it was still a draft at that point, but it was still a deliberate choice)   
   was changed to very bad undefined behavior.  The requirement that p.reset()   
   take a *non-owned* pointer shows that it was already undefined behavior.   
      
      
   --   
         [ See http://www.gotw.ca/resources/clcm.htm for info about ]   
         [ comp.lang.c++.moderated.    First time posters: Do this! ]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)