Re: [boost] [GIL] integration between boost libraries (gil / multiarray)

Subject: Re: [boost] [GIL] integration between boost libraries (gil / multiarray)
From: Christian Henning (chhenning_at_[hidden])
Date: 2010-06-23 10:46:34

Hi there,

> It's not a good thing that some computation functions become dependant to gil only for a point class.
> To compute a geometric transformation, I cannot directly get a gil point, multiply it by a scalar and return it, because this operator is not define in gil...
> Or more generally I cannot directly compute a product with an uBlas matrix. Maybe some cast operators and constructors from/to bounded_vector can be suffiscient to have interoperability.

You can have your own point structure or create some adapters for
yourself. Have a look here:

http://stlab.adobe.com/gil/html/structboost_1_1gil_1_1_point2_d_concept.html#_details

>>In theory you're right. But both libraries were developed by different
>>people at different times without having the need to have compatible
>>concepts.
>
> Yes I understand, I discuss possible improvements. I think there is a need to have compatible concepts.

I can second that. Unfortunately my time frame right now is very
small. Do you have any specific ideas?

>
> A simple solution can be to implement those gil ideas, which can be applied to N-dimensional, in multi-array to propose the same fonctionnalities ?

There is a need for 3D regarding volumetric data. Not sure for more
than that. A general c++ n-dimensional lib is very very hard if not
impossible when performance plays a major role. I'm not aware of a
such a C lib either.

Regards,
Christian