"Johan Nilsson" <johan.nilsson@[...].se>  wrote in message
news:c97cum$cn2$1@[...]..
> 
>  "Jeff Flinn" <TriumphSprint2000@[...].com> wrote in message
>  news:c97afd$60h$1@[...]..
>  >
>  > "Johan Nilsson" <johan.nilsson@[...].se> wrote in message
>  > news:c96p5d$qlg$1@[...]..
>  > >
> 
>  [...]
> 
>  > matter
>  > > if this is a two-step procedure:
>  > > Allocate in constructors + Initialize in entry
>  > > or one-step:
>  > > Allocate and initialize in constructor.
>  >
>  > class astate : non_copyable
>  > {
>  >     T& mRef;
>  >
>  >     astate();
>  >
>  > public:
>  >
>  >     astate( T& aRef ):mRef(aRef){ aRef.somefunction(); }
>  >
>  >     void init( T& aRef ){ mRef = aRef; mRef.somefunction(); } // not the
>  > same
>  >
>  > };
> 
>  No, of course not. I meant:
> 
>  // one-step
>  class astate : non_copyable { ...
>      astate( T& aRef ):mRef(aRef){ aRef.somefunction(); }
>  ... };
> 
>  versus:
> 
>  // two-step
>  class astate : non_copyable { ...
>      astate( T& aRef ):mRef(aRef){}
>      void entry(){ mRef.somefunction(); }
>  ... };

This assumes that you have access to the same information at construction
time as at the call to astate::entry (you previously were discussing
init/deinit). I was pointing out that separating out init() does in fact
come with some cost to flexibility.


Jeff Flinn



_______________________________________________
Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost