casting to void instead of using free?

[dayalsoap]

data_free(void *event_data, void *client_data)
{
    (void) client_data;
    free(event_data);
}

What's the purpose of (void)client_data ?? It's already being passed in as a void pointer....

[Jorl17]

I'd say that it is used to avoid a compiler warning about client_data not being used. However, there are much simpler and beautiful approaches to that...

[dayalsoap]

I'd say that it is used to avoid a compiler warning about client_data not being used. However, there are much simpler and beautiful approaches to that...

I see. is there any reason why you'd pass it in if it's not going to be used??

[bithub]

I see. is there any reason why you'd pass it in if it's not going to be used??

It's possible it is implementing the prototype of a function pointer, but in this implementation, only one of the parameters is used.

[brewbuck]

I'd say that it is used to avoid a compiler warning about client_data not being used. However, there are much simpler and beautiful approaches to that...

Such as what? Casting to void is pretty much the standard way of doing it.

[iMalc]

Such as what? Casting to void is pretty much the standard way of doing it.

I've barely ever seen it done that way. 99% of what I see is simply the commenting out of the parameter name. I.e:

data_free(void *event_data, void * /*client_data*/)

[Prelude]

>99% of what I see is simply the commenting out of the parameter name.
Which only works in C++. In C it's a syntax error to have an unnamed parameter in a function definition.

[Jorl17]

Such as what? Casting to void is pretty much the standard way of doing it.

I just let the argument name out. In the case I want to keep it in anyway, I define a macro:

#define UNUSED(x)

Then I would do:

data_free(void *event_data, void * UNUSED(client_data))

I think that's also what Qt4 does.


However:

>99% of what I see is simply the commenting out of the parameter name.
Which only works in C++. In C it's a syntax error to have an unnamed parameter in a function definition.

I see that, I didn't know it! So it seems that it really is one of the most standard ways of doing it. Other than the classical x = x;

Maybe it could be cleaner with a macro...maybe not!

Also, *without* optimization, doesn't the void method produce a nop?

[Prelude]

>Also, *without* optimization, doesn't the void method produce a nop?
It should produce a no-op regardless (though an exceptionally poor compiler may not do so). The cast simply silences a potential warning.

[iMalc]

>99% of what I see is simply the commenting out of the parameter name.
Which only works in C++. In C it's a syntax error to have an unnamed parameter in a function definition.

Well I guess that speaks volumes about which language I use most