Bug 46492

Summary: document.nodesFromRect() needs to be removed from Document.idl
Product: WebKit Reporter: Sam Weinig <sam>
Component: WebCore Misc.Assignee: Antonio Gomes <tonikitoo>
Status: RESOLVED FIXED    
Severity: Normal CC: abarth, ademar, kenneth, manyoso, mihaip, tonikitoo
Priority: P2    
Version: 528+ (Nightly build)   
Hardware: PC   
OS: OS X 10.5   
Bug Depends on:    
Bug Blocks: 44089, 46336    
Attachments:
Description Flags
(committed with r69345, r=kenneth,kling) patch v1 none

Sam Weinig
Reported 2010-09-24 11:48:09 PDT
The document.nodesFromRect() function should never have been added to Document.idl since that makes it a web facing API. We should remove it ASAP and instead add WebKit level API as necessary. For ports that want this functionality exposed, a good place might be on their WebView. (NOTE: if it is determined that there are good use cases for this feature, we can add it back, prefixed with webkit).
Attachments
(committed with r69345, r=kenneth,kling) patch v1 (7.16 KB, patch)
2010-09-25 09:00 PDT, Antonio Gomes 		no flags
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
Antonio Gomes
Comment 1 2010-09-24 12:15:51 PDT
sam, any chance we can get bug 46336 (Qt needs this change), while I work on that one? So right after bug 46336 gets in, I can: (quick action plan) 1) remove the method from the IDL; 2) Skip the layout test relies on it to be in the IDL. (once _1_ and _2_ are done right way I can:) 3) make it testable through LayoutTestController and unskip the test again. What do you think?
Antonio Gomes
Comment 2 2010-09-25 09:00:22 PDT
Created attachment 68827 [details] (committed with r69345, r=kenneth,kling) patch v1 Patch removes nodesFromRect() from Document.idl, skips the test relying on it, and fixes the window-properties-expected.txt accordingly.
Kenneth Rohde Christiansen
Comment 3 2010-09-25 09:04:26 PDT
Comment on attachment 68827 [details] (committed with r69345, r=kenneth,kling) patch v1 View in context: https://bugs.webkit.org/attachment.cgi?id=68827&action=review > LayoutTests/ChangeLog:9 > + Skip the tests relying on document.nodesFromRect in all platform for > + the moment, and adjusting the expected results tests also accordingly. Are you going to readd these tests later with DRT support?
Antonio Gomes
Comment 4 2010-09-25 09:10:11 PDT
(In reply to comment #3) > (From update of attachment 68827 [details]) > View in context: https://bugs.webkit.org/attachment.cgi?id=68827&action=review > > > LayoutTests/ChangeLog:9 > > + Skip the tests relying on document.nodesFromRect in all platform for > > + the moment, and adjusting the expected results tests also accordingly. > > Are you going to readd these tests later with DRT support? Yes, it will be LayoutTestController::nodesFromRect(document, x, y, topPadding, rightPading, ...). Filing bugs for each platform now. Thanks.
Antonio Gomes
Comment 5 2010-09-26 12:06:55 PDT
Comment on attachment 68827 [details] (committed with r69345, r=kenneth,kling) patch v1 Clearing review flag on attachment: 68827. Committed r68345: <http://trac.webkit.org/changeset/68345>
Adam Barth
Comment 6 2010-09-26 12:10:28 PDT
Oh, nice idea. Can we add that to webkit-patch?
Antonio Gomes
Comment 7 2010-09-26 12:37:16 PDT
(In reply to comment #6) > Oh, nice idea. Can we add that to webkit-patch? That one of the main reasons I am not using webkit-patch to land patches, and we even discussed that briefly in the past: http://www.mail-archive.com/webkit-dev@lists.webkit.org/msg10090.html , including you pointing out where to look at in the code :)
Ademar Reis
Comment 8 2010-11-12 12:27:54 PST
Revision r68345 cherry-picked into qtwebkit-2.1 with commit 659ae6b <http://gitorious.org/webkit/qtwebkit/commit/659ae6b>
Note You need to log in before you can comment on or make changes to this bug.