Join GitHub today
GitHub is home to over 50 million developers working together to host and review code, manage projects, and build software together.
Sign upFixed edge mask decoding in the shader #2254
Conversation
|
There was already a reftest for this (rotated-clip) but it was so small in size that it went under the segment split threshold and rendered in one go, hence it didn't reveal the problem. This PR includes an updated reftest. |
|
The change looks good, thanks! The reftest image is cropped through - I guess it should probably be enlarged to include the entire primitive? Would it also make sense to have the reftest include both a small and large primitive, to test both cases? We should probably do a try run for this too, I guess. |
|
Oh, CI is failing on TC too - might be a slight discrepancy in the reftest output. |
|
Could this also be the cause/fix for #2252 ? |
|
Try looks green: https://treeherder.mozilla.org/#/jobs?repo=try&revision=68770f9175f0c8e963521ab0a775544564898f5c , if you don't mind 4 unexpected passes. The new changes from @bors-servo r=glennw |
|
|
Fixed edge mask decoding in the shader Related to #1797 Apparently, the old GLSL code that I had was all-or-nothing. For some reason, it didn't show the issue before, but it started showing it when the new brush splitting has arrived. <!-- Reviewable:start --> --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/webrender/2254) <!-- Reviewable:end -->
|
|
|
old good OSX timeout... |
Fixed edge mask decoding in the shader Related to #1797 Apparently, the old GLSL code that I had was all-or-nothing. For some reason, it didn't show the issue before, but it started showing it when the new brush splitting has arrived. <!-- Reviewable:start --> --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/webrender/2254) <!-- Reviewable:end -->
|
|
|
ugh, all sorts of Travis failures... to check out the source, to wait for the completion, etc. |
|
@bors-servo retry |
Fixed edge mask decoding in the shader Related to #1797 Apparently, the old GLSL code that I had was all-or-nothing. For some reason, it didn't show the issue before, but it started showing it when the new brush splitting has arrived. <!-- Reviewable:start --> --- This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/webrender/2254) <!-- Reviewable:end -->
|
|
kvark commentedJan 3, 2018
•
edited by larsbergstrom
Related to #1797
Apparently, the old GLSL code that I had was all-or-nothing. For some reason, it didn't show the issue before, but it started showing it when the new brush splitting has arrived.
This change is