[cairo] Review of recent negative-stride changes

[Vladimir Vukicevic]

On Mar 6, 2008, at 5:42 AM, Behdad Esfahbod wrote:

> On Wed, 2008-03-05 at 17:01 -0500, Carl Worth wrote:
>>
>>>    Check that we also allow surfaces to be created using a negative
>> stride.
>>
>> Uhm... is this even something we want to enshrine in the test
>> suite. I certainly don't expect cairo_image_surface_create_for_data  
>> to
>> work with a negative stride. And it's not currently documented that
>> way.
>>
>> Is there some justification for supporting this? If so, then let's
>> change the documentation and ensure the implementation matches. But
>> we're a little late in 1.5.x for a change like that.
>
> Negative strides are useful for integrating with systems having the
> origin at bottom left.  Not sure there are any that we care  
> about...  We
> had some quartz issue like that though.  What was that Vlad?

Quartz uses the PS/PDF coordinate space (origin in bottom left), but  
images always start at the top left, so no issue there.  Win32 can use  
either, though its "native" is bottom-up.  OpenGL however wants bottom- 
up, though you can configure things so that textures get loaded top- 
down.  I don't think it's worth keeping around the support if it's  
confusing, but if it's essentially free (though the maintenance cost  
might not be), we may want to keep it.

     - Vlad