[Insight-users] WrapITK python bindings and GradientAnisotropicDiffusion with 3D float input

[Kalle Pahajoki]

I only noticed this recently since I haven't touched the segmentation 
code in a while.

Gaetan Lehmann wrote:

>
> Kalle,
>
> We (ok, Zach have done most of the work :-)) have now splitted the 
> WrapITK  modules in 27 modules, so your code will no more work.

Sounds good.

> Zach have also introduced a new feature: the modules are not loaded 
> when  we import itk, but when we use a class in that module. The 
> result is that  itk automatically load what you need and nothing more, 
> so you don't have  to care about loading manually such or such module, 
> and the loading time  is a lot shorter :-)
> All the tests are ok, and it works perfectly for me.

Excellent!

>
> Can you update your WrapITK copy, and tell us if it gives good results 
> for  you ?

Everything seems to be working very well. Porting the relatively few 
pieces of code using ITK code was a breeze and I love the lazy loading! 
It made it possible to just throw away the previously ugly code that 
dealt with the separate modules. 

Loading the classes is _a lot_ faster now, up to a level where users can 
easily put up with the wait ;-) And I don't need to bite my nails for 
half a minute everytime I need to re-start the program during development!

Actually there's almost no point in replying to this almost a month 
late, other than I wish to say a big thank you for your work!

Kalle