| Summary: | [GTK] fast/canvas/DrawImageSinglePixelStretch.html fails | ||||||
|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Philippe Normand <pnormand> | ||||
| Component: | Canvas | Assignee: | Martin Robinson <mrobinson> | ||||
| Status: | RESOLVED FIXED | ||||||
| Severity: | Normal | CC: | d-r, krit, leandro, mdelaney7, mrobinson, nishant.646, nishant.bansal, rakuco, webkit.review.bot, zan | ||||
| Priority: | P2 | Keywords: | Gtk, LayoutTestFailure | ||||
| Version: | 528+ (Nightly build) | ||||||
| Hardware: | PC | ||||||
| OS: | OS X 10.5 | ||||||
| Attachments: |
|
||||||
|
Description
Philippe Normand
2011-04-12 02:15:55 PDT
Yea, that's the incorrect behavior you're seeing then. It does show a 'FAILED', correct? I've interpreted this portion of the spec as the inspiration for this test: "If the original image data is a bitmap image, the value painted at a point in the destination rectangle is computed by filtering the original image data. The user agent may use any filtering algorithm (for example bilinear interpolation or nearest-neighbor). When the filtering algorithm requires a pixel value from outside the original image data, it must instead use the value from the nearest edge pixel. (That is, the filter uses 'clamp-to-edge' behavior.)" It also came to my attention that many rendering platforms don't usually do this behavior of clamp to edge because it's not desired in cases where tiling of an image happens and such. Though, the pixels can be first copied into a buffer image and then drawn to effectively create the clamp to edge behavior. The same issue is also present in EFL port. Can you upload and expected and actual image for comparison please? Created attachment 189642 [details]
Patch
(In reply to comment #3) > Can you upload and expected and actual image for comparison please? The test is a text dump, but with a bit of tinkering I was able to see the canvas. It's clear that the issue is that we are sampling outside the source rectangle. I've uploaded a patch fixing the issue. I tried this patch. The issue is still present in EFL port. (In reply to comment #6) > I tried this patch. The issue is still present in EFL port. What version of Cairo? Also, are you using a backend other than image? (In reply to comment #7) > (In reply to comment #6) > > I tried this patch. The issue is still present in EFL port. > > What version of Cairo? Also, are you using a backend other than image? cairo version is 1.12.4. I didn't understand your second query. There was some build problem or something. I tried this patch on the latest webkit and it worked fine. (In reply to comment #9) > There was some build problem or something. I tried this patch on the latest webkit and it worked fine. Thanks for the information. Dirk, maybe you would be willing to review this patch? Comment on attachment 189642 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=189642&action=review LGTM! > Source/WebCore/platform/graphics/cairo/PlatformContextCairo.cpp:196 > + // The pattern transformation properly scales the pattern for when the source rectangle is a Nit, "for when" -> "when" Comment on attachment 189642 [details] Patch Clearing flags on attachment: 189642 Committed r148681: <http://trac.webkit.org/changeset/148681> All reviewed patches have been landed. Closing bug. |