why is it that when you want to assign a function pointer to a function, you have to use the address of operator ( & ) unlike in C/C++. This kinda annoys me because im at university where they force me to use C, and im always having to think to myself "should i put an ampersand here or not?". At the very least make it optional.

On Fri, 06 Feb 2004 13:12:00 +0000, imr1984 wrote:

> why is it that when you want to assign a function pointer to a function, you have to use the address of operator ( & ) unlike in C/C++.

Have you read http://www.digitalmars.com/d/property.html, [cited 06.02.04] about functions as properties?

And, I think I do not grep you fully: do you really mean assigning a function pointer to a function, or the other way round?

So long.

imr1984 wrote:

> why is it that when you want to assign a function pointer to a function, you
> have to use the address of operator ( & ) unlike in C/C++. This kinda annoys me
> because im at university where they force me to use C, and im always having to
> think to myself "should i put an ampersand here or not?". At the very least make
> it optional.
> 

The rule is really simple in D: You *always* need the ampersand when taking the address of something.

 -- andy

imr1984 wrote:
> why is it that when you want to assign a function pointer to a function, you
> have to use the address of operator ( & ) unlike in C/C++. This kinda annoys me
> because im at university where they force me to use C, and im always having to
> think to myself "should i put an ampersand here or not?". At the very least make
> it optional.

Don't gesitate to put an ampersand (&) in C. It works just as well. I find that it's a point of inconsistency in C that you can use both adress-of and a function name itself as if it was the same. If you put & in C you don't get adress-of-adress, it's simply the same as without in context of function pointers. You have to put yet another & to get adress-of-adress IIRC. (btw, don't do this adress of pointer trick unless you know where this adress is stored! or don't do it at all!). Finally, both C++ and D have a more elegant replacement to function pointers, which is virtual methods of classes.

In C i got caught a few times. I forgot to put () after a function call without parameters. And since this yuilds a function adress, which is a value, which is discarded since nothing sensible is done with it, the function doesn't get called. Nothing happens. The program just runs further. I tried to trace into the function with a debugger for hours, till i finally saw that i left out ()! :> So i guess it would be good that it stays an error to assign a function to function pointer.

It happened a few times while writing my first C prog - i was a more or less experienced Delphi developer back then and was really frustrated by endless bugs caused by C - and then when using BCX which generated wrong code of this sort under some circumstance.

-eye

PS. tomorrow i have an exam - wish me luck

imr1984 wrote:

> why is it that when you want to assign a function pointer to a function, you
> have to use the address of operator ( & ) unlike in C/C++. This kinda annoys me
> because im at university where they force me to use C, and im always having to
> think to myself "should i put an ampersand here or not?". At the very least make
> it optional.

In C/C++ the address-of operator is implied for C-style functions so use of it is optional.  It's not implied for member function pointers so use there is mandatory AFAIK.  I like that in D they enforce a degree of consistency here.  But that also means you're free to always use the '&' in C as well.


Sean

"imr1984" <imr1984_member@pathlink.com> wrote in message news:c003r0$l5v$1@digitaldaemon.com...
> why is it that when you want to assign a function pointer to a function,
you
> have to use the address of operator ( & ) unlike in C/C++. This kinda
annoys me
> because im at university where they force me to use C, and im always
having to
> think to myself "should i put an ampersand here or not?". At the very
least make
> it optional.

It's mandatory in D to eliminate parsing ambiguities with respect to properties.