Looks fine.
On 11/4/19 8:29 am, Prasanta Sadhukhan wrote:
Hi All,
Please review a test fix for an test issue seen primarily in linux.
The test draws "AOB" with and without underline underneath and then looks to try to
calculate the midpoint of "O" and then expects to see the transition
from top to bottom of bg->fg->bg->fg->bg as you get when you cross the path of the O twice. When
there is underline, it expects 6 transition from top of O till it passes the underline in form of
bg->fg->bg->fg->bg->fg->bg
as it apparently assumes that there is at least a pixel gap between lower
curvature of O to underline. This may not be true for all fonts, so sometimes
underline are stuck with the lower part of O and it fails.
Proposed fix is to calculate the transition from 10 pixel left from midpoint of
"O" so that there is at least some pixels gap between curvature of O and
underline.
Also, the text is drawn in aliased mode to avoid the antialias zigzag to be mistakenly
construed as "transition".
Proposed fix pass in all platforms and the mach5 job link is present in JBS.
Bug:https://bugs.openjdk.java.net/browse/JDK-8229935
webrev: http://cr.openjdk.java.net/~psadhukhan/8229935/webrev.0/
Regards
Prasanta
--
Best regards, Sergey.
