[cairo] cairo_matrix_t

[Bill Spitzak]

On Tuesday 07 September 2004 11:31 am, Keith Packard wrote:
> Around 9 o'clock on Sep 7, Bill Spitzak wrote:
> > The only good reason for the current implementation is for thread safety.
>
> The other reason is that we can extend the matrix specification without
> affecting existing programs.  It would be nice to support projective
> transformations at some point, but we don't know how to stroke curves under
> such a transformation yet.

It could be defined from the start as being a 4x4 matrix, and indicate that 
certain entries are ignored and must be 1 or 0 for future compatability. Or 
the 4x4 matrix could be a different type with new calls that take it.

I don't think anybody should assumme that getting the matrix and then putting 
it back is a substitute for gsave/grestore.