[cairo] Idle thoughts on error propagation

[Vladimir Vukicevic]

On May 9, 2008, at 7:01 AM, Behdad Esfahbod wrote:
> On Fri, 2008-05-09 at 08:38 +0100, Chris Wilson wrote:
>>
>> First, we have the long-lived surfaces that, for example, are bound  
>> to a
>> window and thus are used repeatedly as the target of multiple  
>> cairo_t.
>> As this is a destination, we don't care about error propagation and
>> having to rebuild the surface would be a hindrance (and error-prone).
>
> I'm still yet to see what kind of errors people have in mind when they
> say the error should not be propagated to the surface.  And no, we  
> don't
> typically have long-lived surfaces bound to a window.  GTK+ creates a
> surface on each expose for example (IIRC).

We will have long-lived surfaces in Gecko in the near future (to keep  
around a permanent backbuffer, among other things).

     - Vlad