[cairo] invalid size error, nth ping

[Behdad Esfahbod]

[You dropped the cairo list CC it seems.  Adding.]

Paolo Bonzini wrote:
>> I didn't look into it in detail, but from the glance I think the idea and API
>> was quite out of place.  Cairo's focus is in API usability.
> 
> All I add publicly is a getter, a setter, and an enum.

Yep, that's what I remember.

>  This is IMO a
> cleaner solution than the cairo_quartz_image_surface that 1) is all
> about caching; 2) my mechanism subsumes and generalizes (so that you
> could use the same client code for Quartz and xlib/xshm for example).
> You could have all surfaces cacheable by default, but it might be a
> waste of resources and it would force to implement reclamation
> policies.

Indeed, that's why the quartz image surface is not supported and due to be
removed/changed/... when we figure out how to implement such functionality.
It's a pity that it made its name into cairo.h.  Then again, I did all I could
do to implement cairo_surface_type_t differently but failed :).


behdad

>> Users should not
>> have to care about caching, let alone calling specific cairo API for it. Your
>> proposed API looked more suited to something like AGG, not cairo.
> 
> Maybe you're confusing the internal and public APIs I was proposing?
> (I'm not saying either of them is perfect, but the comment seems more
> in place if it refers to the internal API).
> 
> Paolo
>