[cairo] The right approach to projective transformations

[Maarten Bosmans]

2010/8/31 Behdad Esfahbod <behdad.esfahbod at gmail.com>:
> On 08/31/10 13:52, Bill Spitzak wrote:
>> Behdad Esfahbod wrote:
>>
>>> Right.  But we should be careful not adding yet another space.
>>> cairo_user_to_device_distance() will suffer no matter what.
>>
>> I think this will have to return the device distance between the
>> transforms of 0,0 and x,y.
>>
>> Or maybe the current point and the current point plus x,y?
>
> I kinda like that.

Yes, but it should be clear from the documentation that once the
projective parts of the transformation matrix are non-zero the notion
of translational invariance is gone and the concept of distance is
dependent on where you are on the surface.

Maarten