[cairo] Regression in stroker

[Vladimir Vukicevic]

On Mar 11, 2008, at 3:57 AM, Adrian Johnson wrote:

> Adrian Johnson wrote:
>> The attached PDF when viewed with the latest git poppler and cairo  
>> has a
>> round cap missing and one of the round joins has been clipped.
>>
>> According to git bisect the regression was introduced by the cairo  
>> commit:
>>
>>  afbd82671fe5ebebe5d58bef3d372312be1c5aeb
>>  "Optimize stroker by hoisting some calculations out of loops"
>
> Attached is a test case that demonstrates the bug. All non horizontal
> lines are missing the round caps. One of the round joins has been  
> flattened.

My bad.  A wrong slope was being set as the device-space slope (the  
user-space slope, in particular); the attached patch passes things  
down correctly.  This also seems to be what caused the slight  
rendering differences with my patch that I attributed to numerical  
precision due to computations happening in different places than  
before.  If this looks ok, I'll commit with a test and updated images.

      - Vlad

-------------- next part --------------
A non-text attachment was scrubbed...
Name: dev-slope-fix.patch
Type: application/octet-stream
Size: 6762 bytes
Desc: not available
Url : http://lists.cairographics.org/archives/cairo/attachments/20080312/93c66659/attachment-0001.obj 
-------------- next part --------------