[cairo] [RFC] Color space API (partial proposal)

[Bill Spitzak]

James Cloos wrote:

> Since it takes several API calls to build up a gradient, I suspect that
> cairo should demand that the colour stops in a given gradient all be in
> the same colour space, thus the API for specifying the alternate colour
> should not need to specify the alternate's space explicitly.

I would make it match the normal gradient.

The gradient itself has a color space (the "blending" space). When you 
specify the stops in it, you can specify a different color space with 
each color. These colors are immediately translated to the blending 
space and then the gradient is a linear interpolation of the stops in 
the gradient's blending space.

When the spot color is drawn a backend that does not understand the spot 
color then mixes the colors from the gradient. This is done the same way 
you would render a point from a normal gradient: the color from the 
gradient is translated from the gradient's blending space to the final 
blending space and mixed in the final space.

I'm sure in reality a lot of these color spaces will be identical, but 
this makes the api consistent.

PS: I also think that whatever pdf wants is going to have to be catered 
to. If an api is made that cannot be trivially translated to some pdf 
api then this is not going to work for the majority of users. This does 
not mean that 100% of pdf needs to be implemented, there are probably 
many ways of specifying spot colors and only one needs to be supported.