From: 0xcdcdcdcd@gmx.at   
      
   On 06.09.2013 18:35, Jean-Marc Bourguet wrote:   
   > Martin Ba <0xcdcdcdcd@gmx.at> writes:   
   >   
   >> On 04.09.2013 21:49, Jean-Marc Bourguet wrote:   
   >>> Martin Ba <0xcdcdcdcd@gmx.at> writes:   
   >>>   
   >>>>> ... disengaged optional is simply treated as an   
   >>>>> additional and unique value of T equal only to itself;   
   >>>>> this value is always compared as less than any   
   >>>>> value of T ...   
   >>>>   
   >>>> - seems awful.   
   >>>   
   >>> If you look at optional as an array of size 0 or 1, it makes   
   >>> perfectly sense.   
   >>>   
   >>   
   >> Actually, no. How does it compare to arrays? Please explain?   
   >   
   > As all other containers, a lexicographic sort.   
   >   
   >> Note: I thought about optional being a model of container where   
   >> the container can at most store one element of T (or be empty). Did you   
   >> mean this by "array"? I don't see how current optional models this?   
   >   
   > AFAIK, it doesn't provide the required interface of containers.  But it   
   > is one of the mental model which has been used for design the interface   
   > of optional<>.   
   >   
      
   I still don't get how what optional does at the moment with regard to   
   the (comparison) operators has got anything to do with what a 1-size   
   container would "naturally" provide.   
      
   Optional allows:   
      
   std::optional<...> theValue;   
   ...   
   if( theValue < MAX_ALLOWED )   
   {   
      // ...   
   }   
      
   That is, it allows comparing the "container"-type to it's underlying type.   
      
   No container I know of does that.   
      
   std::vector<...> theValue;   
   ...   
   assert(theValue.size() == 1);   
   if( theValue < MAX_ALLOWED ) { ...   
      
   ... that doesn't make any sense.   
      
      
   I feel it's a very unfortunate design decision that the (comparison)   
   operators try to save you the dereferencing.   
      
   It may have it's uses in some generic code, but it IMHO bad for all   
   other code.   
      
   Note: I'm fine with the contextual conversion to bool, I don't see much   
   harm in that.   
      
   cheers,   
   Martin   
      
      
   --   
         [ 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)