5fb96358   
   From: moreno_news@mailinator.com   
      
   On 11-11-09 01:31 AM, Joshua Maurice wrote:   
      
   >> Wait a second --- notice that I said I'm trying to convince   
   >> someone that the trick invokes undefined behaviour;  but the   
   >> thing is, his argument is that he got the idea from C++ books   
   >> (not sure which one(s), as I don't recall seeing the technique   
   >> in books),   
   >   
   > Those are some mighty bad C++ books then. Unfortunately, bad C++ books   
   > do exist.   
      
   *sigh*  tell me about it  (long story --- suffice it to say that   
   the poor quality of books, and even when only seen as the   
   *educational* quality of books, has been a pet peeve of mine   
   for quite a long time!)   
      
   >> and it simply *works* on every existing compiler  (well, I should   
   >> probably add that it is likely that it works on every existing   
   >> compiler).   
   >  [ ... ]   
   >> So, it does make sense that he invited/challenged me to show   
   >> him a reputable reference that states that the technique is   
   >> not legal.  (I mean, we have to agree that those aspects in   
   >> the above paragraph, put together, constitute quite solid   
   >> evidence in favor of using the technique).   
   >   
   > Not really, no. That again misses the entire point. The point of   
   > programming to contracts and standards is that we all agree on what is   
   > to be guaranteed now and in the future [ ... ]   
      
   I'm not disagreeing with that --- and for what is worth, the part   
   that "it simply works" was added by me  (it was not one of his   
   arguments).   His main argument is that he found the technique   
   in *two* different books, and he did find that the code was   
   simpler  (I do agree, but I won't necessarily defend that idea   
   too strongly).   
      
   But what I'm saying is that those elements *combined*  (you're   
   focusing too much on the issue that "it works today", as if that   
   was the only reason, and as if we were specifically ignoring an   
   explicit label of UB just because we observe that it works when   
   we try).  In the abscence of knowledge or certainty about whether   
   that is UB, you see the technique in two separate books, you try   
   it, it works, I think it is more than reasonable to assume that   
   it works because it is supposed to work, and not by mere   
   coincidence  (as it is the case).   
      
   Cheers,   
      
   Carlos   
   --   
      
         [ 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)