Summary: | Fix for drawing invalid layers in WebViewBenchmarkSupportImpl | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Daniel Murphy <dmurph> | ||||
Component: | New Bugs | Assignee: | Daniel Murphy <dmurph> | ||||
Status: | RESOLVED FIXED | ||||||
Severity: | Normal | CC: | alokp, crogers, enne, jamesr, nduca, scheib, webkit.review.bot | ||||
Priority: | P2 | ||||||
Version: | 528+ (Nightly build) | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Attachments: |
|
Description
Daniel Murphy
2012-08-01 16:12:55 PDT
Created attachment 155916 [details]
Patch
What's the bug that this fixes? Comment on attachment 155916 [details]
Patch
Are there any tests for this feature?
There was an assertion being thrown when we tried to paint non-content layer, at RenderLayerBacking.cpp:1219. This only shows up on websites with backing layers like the poster circle. No, there are none. Testing this specifically seems a bit infeasible. This provides the drawing for canvas serialization, so when that's working we can do bitmap based tests. I think it will be nice to have a simple webkit test for WebViewBenchmarkSupport::paint() since it uses a path different from the actual paint path. Testing it downstream will be hard to maintain. I think no test is needed for this. This is not because I think testing is unnecessary. But rather, that testing for this specific bug is busy work. There is a broader question of how to test the support code. And whether its testable. I propose we land this fix, and then Alokp, Daniel and I deal with the testability discussion offline. It may be that we're going to change this more heavily. (In reply to comment #7) > This is not because I think testing is unnecessary. But rather, that testing for this specific bug is busy work. There is a broader question of how to test the support code. And whether its testable. I think that was really my question. I hadn't ever seen WebViewBenchmarkSupport before and was curious if there was any testing strategy here. > I propose we land this fix, and then Alokp, Daniel and I deal with the testability discussion offline. It may be that we're going to change this more heavily. Sure. Comment on attachment 155916 [details]
Patch
R=me.
Comment on attachment 155916 [details] Patch Clearing flags on attachment: 155916 Committed r124391: <http://trac.webkit.org/changeset/124391> All reviewed patches have been landed. Closing bug. |