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 upBetter inversible transformations #2913
Conversation
|
Looks good to me (with the extra commits removed). To test it you can add this patch on top of this revision https://treeherder.mozilla.org/#/jobs?repo=try&revision=04c6bc71074245fa9d8438a78922e71fbd36dff7&selectedJob=188748580 and run the crashtests locally |
|
@bors-servo r+ |
|
|
|
I started a try push that includes #2912 and this PR: https://treeherder.mozilla.org/#/jobs?repo=try&revision=3623dff5d67e5c5a5c77fb5f7a6ed55f7f69cd06 |
Better inversible transformations Fixes #2911 (how can I verify?) r? @nical <!-- 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/2913) <!-- Reviewable:end -->
|
This patch appears to be insufficient, we're still getting panics (see try push above). It's hitting the |
|
@staktrace this makes sense. I'm still working (closest WR bug is #2272) on that part of code (just switching on and off, mostly off, occasionally), and I know it's not yet entirely correct, so unwrapping does no good there (until it's fixed). How about a temporary workaround in 14ebffb ? |
|
Started a new try push with that: https://treeherder.mozilla.org/#/jobs?repo=try&revision=78d838b9b1f8b0d9e014fe5f2ee05f53135aa2e9 |
|
Above try push still had failures, this time from this line per my local repro on |
|
@staktrace thanks for tracing this crash down! I updated the second commit here to somewhat work around it. Note: the workaround is super unsound and in normal circumstances shouldn't be merged. However, the old code was essentially producing undefined values (behind the near plane) by that transformation, so defaulting to concrete ones, while still wrong, is a step forward. |
|
I did a local build with that patch (try still going here) and it crashes still. Same test as last time, but a different spot:
webrender/webrender/src/batch.rs Line 1717 in 2cb6825 |
|
@bors-servo r+ |
|
|
Better inversible transformations Hopefully addresses #2911 (how can I verify?) r? @nical <!-- 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/2913) <!-- Reviewable:end -->
|
|
|
@jrmuizel the errors in this try push appear from #2891:
|
Proper near plane splitting The PR is built on the shoulders of #2913, #2741, servo/euclid#277, servo/euclid#291, servo/plane-split#12, and (last but not the least!) servo/plane-split#15 It uses the new clipping semantics in `plane-split` crate uniformly for both bounding box computation and 3d plane splitting itself, ensuring that no incorrect perspective divisions are performed (and closing quite a few TODO entries). Also adds a small reftest for the latter. Fixes #2272 <!-- 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/2947) <!-- Reviewable:end -->
kvark commentedJul 19, 2018
•
edited by nical
Hopefully addresses #2911 (how can I verify?)
r? @nical
This change is