[Insight-developers] Feature Bloat

[Williams, Norman K]

Thanks for singling me out ;-)


As for feature bloat, someone in our group needed this new capability, and
Hans (who is my boss) asked me to implement it.  So the larger question of
feature bloat in this particular case is literally above my pay grade.

There are two alternatives to the way I implemented this change: 1) Write
some bespoke code for the researcher in our group who needed this feature,
and don't bother sharing it with the larger community and 2) Add yet
another set of Image Math classes to implement physical space processing.
Neither of those options seem better to me than what I implemented.

I don't understand your comment 'does not adjust the requested region
base[d] on the needed regions.'  Not that you're wrong -- you're probably
right -- but I don't  know how to implement that adjustment.

The implementation works using interpolators on the 2nd (and 3rd) images;
I can't begin to know how I'd deal with streaming in those cases.  The
interpolator doesn't care about requested regions as far as I know -- you
give it a physical point and it returns an interpolated pixel at that
point.

If you can explain what change would have to take place, I'll gladly try
to code it.

The way it interacts with the constant value data object as inputs is that
it doesn't. If one or the other of the inputs is a constant, it works in
index space.

On 6/1/11 9:50 AM, "Bradley Lowekamp" <blowekamp at mail.nih.gov> wrote:

>Hello,
>
>I have some concerns that many basic filters are quickly becoming feature
>bloated. There have been many patches along these lines this is just one
>example.
>
>I just saw the following patch:
>http://itk.org/gitweb?p=ITK.git;a=commitdiff;h=cee7ada7339cba22bc154cede71
>16346b13c3705
>
>This new feature does not adjust the requested region bases on the needed
>regions for the operations.  This means it will not would with streaming,
>and may produce undefined behavior (potentially segfault ). Additionally,
>it's not clear to me how this interacts with Gaetan's new constant value
>data object as an input.
>
>I am seeing a lot of new features being added to filters. And I am
>concerned that there is no larger plan for the features that people want
>to add to filters, and  the consequences of maintaining all these new
>features.
>
>
>Or perhaps I am just getting set in my ways because I am working from my
>basement for a little bit today :)
>
>Brad



________________________________
Notice: This UI Health Care e-mail (including attachments) is covered by the Electronic Communications Privacy Act, 18 U.S.C. 2510-2521, is confidential and may be legally privileged.  If you are not the intended recipient, you are hereby notified that any retention, dissemination, distribution, or copying of this communication is strictly prohibited.  Please reply to the sender that you have received the message in error, then delete it.  Thank you.
________________________________