[vtk-developers] Templated filters

[David Gobbi]

Hi Luis,

Thank you for the apology.  Your opinions were already clear in the
origin email, though, so I don't understand why you feel that you have
to repeat yourself.

   David


On Tue, Feb 9, 2010 at 7:32 AM, Luis Ibanez <luis.ibanez at kitware.com> wrote:
> On Tue, Feb 9, 2010 at 3:22 AM, Biddiscombe, John A. <biddisco at cscs.ch> wrote:
>> I'd like to agree with Bill, and possibly Luis's long winded rant (though I'm not sure what he really wanted to say).
>>
> ----------------------------------------------------------------------
>
>
> I must apologize to David and to the list
> for my poorly phrased email of yesterday.
>
>
> It wasn't only rude but it also missed to clearly
> convey the points that I wanted to make.
>
>
> The points are:
>
>
> A) Although ITK and VTK are very similar systems,
>     they have different domains of application.
>
>     Features that may be considered essential in VTK
>     might be undesirable in ITK, and the other way around.
>
>     Evaluating ITK design's decisions based on the
>     requirements of VTK is as misguided as evaluating
>     VTK's design and implementation based on the
>     requirements of ITK.
>
>
> B)  ITK enforces type-awareness in application developers.
>     This is essential when dealing with medical applications.
>
>     It simply comes down to a requirement for enforcing
>     correctness and promoting safety when managing
>     patient data.
>
>     When application developers miss type-awareness they
>     can easily destroy information that is vital for the proper
>     functioning of a medical application.
>
>     Using C++ Templates as a mechanism for enforcing
>     type-awareness is certainly an inconvenience for wrapping,
>     but wrapping is not the guiding consideration in the context
>     of medical applications.
>
>     Wrapping makes life easier for developers who don't
>     want or don't need to be aware of specific types, for
>     example the pixel types of images.
>
>     In the context of medical applications, relaxing type
>     awareness is only acceptable for doing:
>
>                      1) Rapid prototyping
>                      2) Teaching exercises
>
>
>      Medical applications, that go into clinical use, require
>      type-awareness and most likely will be implemented
>      using C++.
>
>
>
> Again, my apologies to David and to the list for my failure
> to express technical ideas in a civilized manner.
>
>
>
>        Luis
>