[vtk-developers] [Paraview-developers] Moving vtkMemberFunctionCommand to VTK

[David Gobbi]

A reinterpret_cast gives an undefined result when casting an object
with multiple inheritance.  So I think that it is better to live with
the overhead of a dynamic cast, just in case someone tries using this
with a class like vtkQtView (it multi-inherits from QObject).

  David


On Wed, Oct 6, 2010 at 9:51 AM, Utkarsh Ayachit
<utkarsh.ayachit at kitware.com> wrote:
> You actually need a reinterpret_cast<>.
>
> Utkarsh
>
> On Wed, Oct 6, 2010 at 11:45 AM, David Gobbi <david.gobbi at gmail.com> wrote:
>> The dynamic_cast is necessary, after all.  The problem is here:
>>
>>  T* handler = static_cast<T*>(this->Handler.GetPointer());
>>
>> The GetPointer() returns a vtkObjectBase, so this line of
>> code refuses to compile for "other" objects.
>>
>>  David
>>
>> On Wed, Oct 6, 2010 at 7:48 AM, Marcus D. Hanwell
>> <marcus.hanwell at kitware.com> wrote:
>>> I was about to ask if a static_cast would work in this case, and it seems
>>> preferable (and equivalent to) a C-style cast. I was just looking at the new
>>> API, and it looks great to me. I think this is a great addition to our C++
>>> API, and am very much in favor of making it slicker where we can.
>>> Marcus
>>> On Wed, Oct 6, 2010 at 9:25 AM, David Gobbi <david.gobbi at gmail.com> wrote:
>>>>
>>>> I was wondering about that dynamic_cast but I thought there
>>>> was a reason for it... but I can change it to a static_cast
>>>> for my commit.
>>>>
>>>> So I'll stage and commit as soon as I can afford a full VTK
>>>> rebuild.
>>>>
>>>>  David
>>>>
>>>> On Wed, Oct 6, 2010 at 7:06 AM, Utkarsh Ayachit
>>>> <utkarsh.ayachit at kitware.com> wrote:
>>>> > I like your changes. In absence of your changes, if a developer does
>>>> > indeed want to achieve same effect he'll be doing something similar
>>>> > and will indeed have the onus of worrying about dangling ptrs. So I
>>>> > don't think this makes it any more confusing. We should probably add
>>>> > documentation to that effect and that pretty much should do it.
>>>> >
>>>> > About the dynamic_cast, yea static_cast should be fine too and totally
>>>> > safe in this case. If the developer was stupid enough to pass in a ptr
>>>> > as a wrong type when making the AddObserver call, he deserves the side
>>>> > effects :).
>>>> >
>>>> > I'd vote for committing your changes. (You should use the stage, add
>>>> > your commit to the "member_function_observers" branch and then merge
>>>> > it in).
>>>> >
>>>> > Utkarsh
>>>> >
>>>> > On Wed, Oct 6, 2010 at 8:24 AM, David Gobbi <david.gobbi at gmail.com>
>>>> > wrote:
>>>> >> On Wed, Oct 6, 2010 at 6:01 AM, David Doria <daviddoria at gmail.com>
>>>> >> wrote:
>>>> >>>
>>>> >>> David G,
>>>> >>>
>>>> >>> Are you suggesting this in place of what Utkarsh pushed? Utkarsh, can
>>>> >>> you take a look and let David G know if he should push it? I think it
>>>> >>> is a good idea to allow event handling from as many places as possible
>>>> >>> (i.e. I support NOT requiring the class to derive from vtkObjectBase).
>>>> >>
>>>> >> I'm just putting it out for discussion.  It does come with some danger,
>>>> >> for example you should not use it instead of vtkEventQtSlotConnect,
>>>> >> because Qt's signal/slot mechanism is safe, and this is not.  It's only
>>>> >> safe when the listener is a vtkObjectBase class.
>>>> >>
>>>> >>  David
>>>> >>
>>>> > _______________________________________________
>>>> > Powered by www.kitware.com
>>>> >
>>>> > Visit other Kitware open-source projects at
>>>> > http://www.kitware.com/opensource/opensource.html
>>>> >
>>>> > Follow this link to subscribe/unsubscribe:
>>>> > http://www.vtk.org/mailman/listinfo/vtk-developers
>>>> >
>>>> >
>>>> _______________________________________________
>>>> Powered by www.kitware.com
>>>>
>>>> Visit other Kitware open-source projects at
>>>> http://www.kitware.com/opensource/opensource.html
>>>>
>>>> Follow this link to subscribe/unsubscribe:
>>>> http://www.vtk.org/mailman/listinfo/vtk-developers
>>>>
>>>
>>>
>>
>