www.digitalmars.com         C & C++   DMDScript  

D - pointers once again (hopefully the last time)

reply "Pavel Minayev" <evilone omen.ru> writes:
Returning to my favourite topic of pointers-to-methods...

Okay, so I understand why you don't want to add them in D. Then, can
you do just one little thing: make it so that unary &, when used on
methods, returns their position in vtable? Using this and the knowledge
of internal structure of Object (I believe it's not "implementation-
defined", right?), I'll do the wrapper myself.

You ask there is a problem with interfaces, well, this could be
solved by only allowing & to be used on methods of class that
aren't implementations of some interface (so they reside in the
very first vtable)...
Jan 02 2002
next sibling parent reply "Pavel Minayev" <evilone omen.ru> writes:
"Pavel Minayev" <evilone omen.ru> wrote in message
news:a0vk3j$25i3$1 digitaldaemon.com...


 you do just one little thing: make it so that unary &, when used on
 methods, returns their position in vtable?


Or, even better, return it as a pointer to global function.
For example, this case:

    class MyForm : Form
    {
        void Button1_OnClick(Object sender) { ... }
    }

Here, &Button1_OnClick would be of type void(*)(Object). This would
allow for better typechecking, like this:

    class ClickEvent
    {
        this(void(*handler)(Object), Object owner) { ... }
    }

    class MouseEvent
    {
        this(void(*handler)(int, int, int, int), Object owner) { ... }
    }

Of course, when it comes to calling the handler, Event takes care
of passing "this" pointer appropriately (does it go first or last BTW?).
Jan 02 2002
parent reply "Walter" <walter digitalmars.com> writes:
"Pavel Minayev" <evilone omen.ru> wrote in message
news:a0vleu$26ha$2 digitaldaemon.com...

 "Pavel Minayev" <evilone omen.ru> wrote in message
 news:a0vk3j$25i3$1 digitaldaemon.com...


 you do just one little thing: make it so that unary &, when used on
 methods, returns their position in vtable?


 Or, even better, return it as a pointer to global function.
 For example, this case:

     class MyForm : Form
     {
         void Button1_OnClick(Object sender) { ... }
     }

 Here, &Button1_OnClick would be of type void(*)(Object). This would
 allow for better typechecking, like this:

     class ClickEvent
     {
         this(void(*handler)(Object), Object owner) { ... }
     }

     class MouseEvent
     {
         this(void(*handler)(int, int, int, int), Object owner) { ... }
     }

 Of course, when it comes to calling the handler, Event takes care
 of passing "this" pointer appropriately (does it go first or last BTW?).


But then you get the pointer to the real function, and it is no longer
virtual. -Walter
Jan 02 2002
parent "Pavel Minayev" <evilone omen.ru> writes:
"Walter" <walter digitalmars.com> wrote in message
news:a0vlsl$26rh$1 digitaldaemon.com...


 But then you get the pointer to the real function, and it is no longer
 virtual. -Walter


Absolutely. But I don't need them to be virtual! I want some way to
bundle a member function to a slot. Who needs to override event
handlers (like Button1_OnClick) anyhow?

Besides, if this is an issue, you could make the pointer actually
be an integer defining the position of function in the vtable...
but the type would still be function pointer. This is only used
for type-checking anyhow...
Jan 02 2002
prev sibling parent reply "Pavel Minayev" <evilone omen.ru> writes:
Could you please state if method pointers (or my latest
suggestion in this thread) are or are not supported in D?
I'm sitting here and don't know what to do - hope to
get this question clearer... if you say "yes", great, if
"no", I'll use a partial solution: delegates. Anything
is better than waiting.
Jan 09 2002
parent "Walter" <walter digitalmars.com> writes:
For an answer right now, it'll be "no". Doesn't mean it will always be that
way, doesn't mean that method pointers are a bad idea, I just am buried with
work and I can't address it now.

"Pavel Minayev" <evilone omen.ru> wrote in message
news:a1hql4$1gae$1 digitaldaemon.com...

 Could you please state if method pointers (or my latest
 suggestion in this thread) are or are not supported in D?
 I'm sitting here and don't know what to do - hope to
 get this question clearer... if you say "yes", great, if
 "no", I'll use a partial solution: delegates. Anything
 is better than waiting.
Jan 09 2002