[cairo] Extents of degenerate paths

[T Rowley]

On 5/10/07 3:09 PM, Behdad Esfahbod wrote:
> On Thu, 2007-05-10 at 12:12 -0500, T Rowley wrote:
>> While I do think that a hairline mode would be useful, in this case all 
>> I wanted to do was measure the extent of a piece of geometry exclusive 
>> of stroke properties.  This is needed in SVG for object bounding box 
>> units and DOM APIs.
> 
> That's the same as fill extents most of the time.  I still think a
> path_extents helps.

Right, in our current code we call fill_extents first, and if that 
returns nothing we retry with stroke_extents.  This did work at some 
point in the past, though we've also changed the stroke width from an 
epsilon to zero at some point, so tracking down exactly when is tricky 
especially given the uneven update schedule of moz's copy of cairo.