[Insight-developers] Test framework

[Hans Johnson]

Bill,

I certainly have a skewed perception here.  My "customers" are the 25
engineering graduate students that I will be teaching this spring that have
just finished taking a course in modern C++/generic programming/object
oriented programming techniques where they learned all the nasty
non-portable little tricks in C++.

The lack of support for partial specialization is a recurrent theme that
burns many new developers.

I like to think of the 100's of developers working hard on improving ITK as
one of the major customers, and wanted to make sure that at least some
mention of our concerns were included in the discussion.



Hans




On 12/30/08 2:23 PM, "Bill Lorensen" <bill.lorensen at gmail.com> wrote:

> Sorry. e-mail sent too soon...
> 
> Hans,
> 
> I think your view is very developer-oriented, not customer-oriented.
> That's OK, I respect your opinion.
> 
> As an academic your priorities may be different than those of a
> company that must support products on a large number of platforms. I
> understand that some day, new algorithms that can only be implemented
> with particular c++ features will require an upgrade.
> 
> I would be willing to look at the new code you mentioned that can only
> be implemented with partial specialization. In the past, the use of
> such code is not necessarily a requirement to implement an algorithm.
> However, I do admit, that in the case of the new gdcm, that partial
> specialization is so prevalent that partial specialization cannot be
> removed. Personally, I think that is unfortunate since reading dicom
> files does not necessarily require partial specialization. Rather,
> this is a convenience for the developer.
> 
> Bill