Affine transforms - matrix algebra

[Martin Desruisseaux]

Hello all

Le 17/08/12 17:56, Pavel Safrata a écrit :
> The methods that take arrays of doubles were requested by users for 
> interoperability with other libraries (e.g. physics) which often 
> support the array conversions.

Just for the record, there is an other use case in favour of arrays. I 
have found the following pattern quite common:

* Get the matrix coefficients as an array.
* Update just a few coefficients in a way that can't be done easily with 
the standard scale/translate methods.
* Set the matrix coefficients to the updated array.

So working with arrays avoid the need to deal with 12 coefficients when 
we want to update only 1 or 2 of them and leave the other unchanged. An 
other rational is that we typically get or set the coefficients using 
some loops, in which case an array is more convenient than individual 
variables.

     Regards,

         Martin