General talk about software patterns

["Christophe Addinquy" <Christophe.Addinquy AT valtech.fr>]

As far as I remember, John Vlissides material was first published in C++ 
Report, then synthitized in the excellent book "Patterns Hatching". I don't 
know if the material is more or less complete, but the advantage of Eric's 
link is the instant access..
 
Regards,
 
 
Christophe

        -------- Message d'origine-------- 
        De: 
patterns-discussion-bounces AT cs.uiuc.edu
 de la part de Eric Y. Theriault 
        Date: sam. 25/09/2004 04:28 
        À: E. Segura 
        Cc: 
patterns-discussion AT cs.uiuc.edu
 
        Objet: Re: [patterns-discussion] deleting a Singleton
        
        

        E. Segura wrote:
        
        >When using the Singleton pattern...
        >
        >Who should delete the _instance?
        >
        >Should I ask for it and delete it myself?
        >( i.e. delete(myClass::getInstance()) ????? )
        >
        >or the class should keep a "reference count" and other
        >classes should inform when the don't use the instance
        >reference anymore.
        > 
        >
        
        John Vlissides has written a great article on the topic, entitled "To
        Kill a Singleton", which I believe you can get at
        http://www.research.ibm.com/designpatterns/pubs/ph-jun96.txt .  Andre
        Alexandrescu's Modern C++ Design book also has a good treatment of the
        topic, where he basically allows this to be specified by a policy.  In
        other words, the users of the singleton object should be (more or 
less)
        oblivious to how it is constructed and destroyed.
        
        I would avoid the delete mechanism above, since this assumes that the
        pointer is allocated by new.  In actuality, the value returned by
        getInstance() is not necessarily allocated by new, as Alexandrescu
        refers to the Meyers Singleton, where the singleton object is a static
        variable.
        
        The above two resources offer are excellent insight to this issue.
        
        
        
        eyt*
        --
        Eric Y. Theriault
        http://www.eyt.ca
        
        _______________________________________________
        patterns-discussion mailing list
        
patterns-discussion AT cs.uiuc.edu
        http://mail.cs.uiuc.edu/mailman/listinfo/patterns-discussion