The layout test editing/inserting/6633727.html can be written using dumpAsText so that the results can be independent of minor platform differences. (Currently it is failing on Chromium/Linux due to subtle font differences; rather than rebaselining we'll want to replace it with dumpAsText based testing.)
Created attachment 43351 [details] Rewrite the test editing/inserting/6633727 to use dumpAsText
The test transformation by itself looks 'equivalent' to me, the render tree dump seem to correspond to html you verify. The dump also has caret information but it seems not needed because the test inserts '2' to verify where insertion point is. I'm not sure I would convert this test to a script test though. It seems like 2 transformations at once - to 'dumpAsText' and to be a script text. It'd be good if Adele took a look, both as author of the test and able to verify what that radar issue is exactly about.
Attachment 43351 [details] passed the style-queue
Comment on attachment 43351 [details] Rewrite the test editing/inserting/6633727 to use dumpAsText From reading the ChangeLog and bug description, it sounds like the bug is at least in part about where the insert position is at the end of this test. That aspect (where the cursor is) is no longer verified after this conversion. It is verified when there is an image because the cursor is visible in that image.
Created attachment 50085 [details] Patch
(In reply to comment #4) > (From update of attachment 43351 [details]) > From reading the ChangeLog and bug description, it sounds like the bug is at > least in part about where the insert position is at the end of this test. That > aspect (where the cursor is) is no longer verified after this conversion. It is > verified when there is an image because the cursor is visible in that image. Thanks for reviewing, I updated the patch to explicitly check the cursor position.
Comment on attachment 50085 [details] Patch It would be nice if the chromium pixel tests results were deleted with this submission (now that they are in the webkit repository).
Comment on attachment 50085 [details] Patch Clearing flags on attachment: 50085 Committed r56506: <http://trac.webkit.org/changeset/56506>
All reviewed patches have been landed. Closing bug.