Re: [boost] [unnullable_ptr] Proposal for a new library

From: Mathias Gaunard (mathias.gaunard_at_[hidden])
Date: 2008-07-22 13:22:06

Andrey Tcherepanov wrote:
> Oh, no... I think the fact that Boost.Function allows storing of null
> pointer ("there is no function to call") is a good thing!

Well, if you like the possibility of invoking undefined behaviour, it is
surely a good thing.
I would rather have the type system and the exception mechanism make
sure that can never happen.

> Otherwise it
> will need to store some sort of "default" function that returns ...
> seriously, what is default function would be? Say we have
> 
>     Func<bool(void)> f;

That line would be an error.
Not allowing function to be empty means no default constructor, obviously.

> To disallow default
> constructor (no argument passes) completely will be way too limiting

I don't really see how.
If you want a function that might or might not be there, you can use
optional<function<sig> >.

> it is a encapsulation of function with pointer to function sort-of
> behaviour, not reference to it.

Yes, the reason Boost.Function behaves that way is to be consistent with
function pointers.
But maybe it could actually try to be better than these and provide
safer guarantees.

> If you completely need to be sure that there is something to call
> underneath it (like to avoid thousand checks for empty value), just
> (wrap and) assign some dummy function.

Making a wrapper is of course possible. It seems fairly weird though to
make a wrapper around an optional container to make it unoptional rather
than the other way around.
A component should have as little features as possible to be complete,
and then one is to combine components to have more refined ones with
more features.

Also, I don't really know how Boost.Function behaves when copy
constructors inside operator= throw, but depending on how it does it it
could lead to a potential empty object. It's not really trivial to
ensure the never-empty guarantee when using stack allocation (SBO).

Putting dummy functions is of course out of the question.