RESOLVED FIXED 11926
<div> tag causes the horizontal scrollbars display 
https://bugs.webkit.org/show_bug.cgi?id=11926
Summary <div> tag causes the horizontal scrollbars display
Madhu M
Reported Friday, December 22, 2006 2:20:30 PM UTC
<html> <body> <div id="div1" style="top:-800;position:absolute;overflow:hidden"><xmp><span id="span1"><table><tr><td>&nbsp;</td></tr><tr><td>Unauthorized Overtime</td></tr><tr><td>Incorrect Assignment</td></tr><tr><td>Error</td></tr><tr><td>IECPP: Incorrect Account Assignment</td></tr><tr><td>Wrong Activity type</td></tr><tr><td >Wrong Order</td></tr><tr><td >Wrong A/A Type</td></tr></table></span></xmp></div> </body> </html> The attached html if opened in Safari causes the Horizontal scroll bar to get displayed. But in Firefox it is not comming. This html is having an <div> tag and an <xmp> inside
Attachments
Sample html showing the bug (479 bytes, text/html)
2006-12-22 06:24 PST, Madhu M 		no flags
Details
patch for this issue (4.41 KB, patch)
2007-03-26 18:28 PDT, Madhu M 		hyatt: review-
DetailsFormatted DiffDiff
patch for this issue (13.99 KB, patch)
2007-04-02 17:51 PDT, Madhu M 		hyatt: review-
DetailsFormatted DiffDiff
Patch to fix this bug. (2.52 KB, patch)
2007-05-10 02:00 PDT, Dave Hyatt 		oliver: review+
DetailsFormatted DiffDiff
RTL case broken after the fix (363 bytes, text/html)
2007-05-10 02:26 PDT, mitz 		no flags
Details
Show Obsolete (2) View All Add attachment proposed patch, testcase, etc.
Madhu M
Comment 1 Friday, December 22, 2006 2:24:00 PM UTC
Created attachment 11963 [details] Sample html showing the bug When the file is opened in Safari it causes the horizontal scroll bar to get displayed. There is nothing on the web page to be displayed, but scroll bar comes. In firefox the behaviour is proper, ie no scroll bar is displayed
David Kilzer (:ddkilzer)
Comment 2 Friday, December 22, 2006 2:42:36 PM UTC
This looks correct to me. The "top:-800;" style attribute of the DIV makes the content of the XMP tag "invisibile", but if you remove that, you'll see that the horizontal scrollbar is needed to view all of the content.
David Kilzer (:ddkilzer)
Comment 3 Friday, December 22, 2006 2:52:28 PM UTC
(In reply to comment #2) > This looks correct to me. The "top:-800;" style attribute of the DIV makes the > content of the XMP tag "invisibile", but if you remove that, you'll see that > the horizontal scrollbar is needed to view all of the content. It also makes it "invisible"!
Jake Logan
Comment 4 Friday, December 22, 2006 7:31:16 PM UTC
(In reply to comment #3) > (In reply to comment #2) > > This looks correct to me. The "top:-800;" style attribute of the DIV makes the > > content of the XMP tag "invisibile", but if you remove that, you'll see that > > the horizontal scrollbar is needed to view all of the content. > > It also makes it "invisible"! > Hi David, I do understand that if you remove the "top:-800;" that will make the content of the DIV visible. However, the purpose of this request is to cause the content of the <XMP> to be *invisible* and to take 0 pixels of space. The XMP tag is used to load code dynamically and the intention is for that code to be invisible and not affect the layout of the DIV at all. We can't change that code and it works as I describe in Firefox and IE. Can you please reconsider? This issue causes erroneous nested scrollbars to appear in Safari in many SAP installations, including some important Apple customers. It is one of the key issues that needs to be fixed for SAP/Safari compatibility. Best, Jake
David Kilzer (:ddkilzer)
Comment 5 Friday, December 22, 2006 8:05:57 PM UTC
> Can you please reconsider? This issue causes erroneous nested scrollbars to > appear in Safari in many SAP installations, including some important Apple > customers. It is one of the key issues that needs to be fixed for SAP/Safari > compatibility. Interesting. I was just reporting what I found in a brief analysis of the HTML. (My CSS analysis needs some work!) After reviewing the CSS, should the "overflow:hidden" property be clipping the content that is not visible, and thus not drawing a scrollbar? Confirming bug--test case works as expected in Firefox 1.5.0.9 (on WinXP).
Jake Logan
Comment 6 Friday, December 22, 2006 8:31:35 PM UTC
(In reply to comment #5) > > Can you please reconsider? This issue causes erroneous nested scrollbars to > > appear in Safari in many SAP installations, including some important Apple > > customers. It is one of the key issues that needs to be fixed for SAP/Safari > > compatibility. > > Interesting. I was just reporting what I found in a brief analysis of the > HTML. (My CSS analysis needs some work!) After reviewing the CSS, should the > "overflow:hidden" property be clipping the content that is not visible, and > thus not drawing a scrollbar? > > Confirming bug--test case works as expected in Firefox 1.5.0.9 (on WinXP). > Yes, I believe that the "overflow:hidden" should cause the content to be clipped (or hidden) and not show scrollbars, as described here: http://www.w3schools.com/css/pr_pos_overflow.asp and here: http://www.quirksmode.org/css/overflow.html I'm no CSS expert either, but the behavior we see in Firefox (both Mac and Windows) I believe to be correct. Thanks for your thoughts. Jake
Dave Hyatt
Comment 7 Friday, December 22, 2006 9:51:19 PM UTC
Yeah this looks like a bug.
Alice Liu
Comment 8 Wednesday, January 31, 2007 11:42:49 PM UTC
<rdar://problem/4694859>
Madhu M
Comment 9 Saturday, March 24, 2007 12:45:14 AM UTC
I have got the reason for this isuue in Webkit. It is while calculating thr right most position in RenderBolck (RenderBlock::rightmostPosition) for positined objects. Here the overflow clip is not considered and the check is done only for the objects having position:Fixed. There is a comment in code like :- " Fixed positioned objects do not scroll and thus should not constitute part of the rightmost position" (?). I think overflow:hidden is to be considered here. When I give check with hasOverflowClip() function, it will ignore that object also while calculationg width and right most position. I think it is a valid fix for the issue. Regards Madhu
Madhu M
Comment 10 Tuesday, March 27, 2007 2:28:22 AM UTC
Created attachment 13823 [details] patch for this issue I am submitting the patch for this issue. The scroll bars are getting displayed because of wrong width calculation. For calculating the width, the right most position is used. But it is not considering the overflow clip, while checking for rightmost position. The patch has a layout test case. This patch is not creating any failures in the existing layout tests.
Dave Hyatt
Comment 11 Tuesday, March 27, 2007 3:11:20 AM UTC
Comment on attachment 13823 [details] patch for this issue This is not the right fix. rightmostPosition should have returned the correct answer. You also didn't patch lowestPosition or leftmostPosition (which presumably suffer from the same issue).
Dave Hyatt
Comment 12 Tuesday, March 27, 2007 3:13:39 AM UTC
Comment on attachment 13823 [details] patch for this issue (The reason this isn't right is you need to at least consider the positioned object's border box, even when overflow is not set to visible.)
Dave Hyatt
Comment 13 Tuesday, March 27, 2007 4:08:17 AM UTC
The real issue here is that content that cannot be scrolled to in one direction (e.g., vertically) should not be allowed to affect a different axis (horizontal).
Dave Hyatt
Comment 14 Tuesday, March 27, 2007 4:09:06 AM UTC
This basically includes content positioned entirely "above" or "to the left" of the scrollable area.
Madhu M
Comment 15 Tuesday, March 27, 2007 11:34:14 PM UTC
Hi Dave Hyatt, Thanks for your comments. From your comments I have made the following points 1. Along with rightmost position we have to make similar checks for lowestPosition or leftmostPosition 2. If the overflow is given as hidden we need to consider the render box of thepositioned objects to ensure that overflow is clipped in both the directions. 3. Contents which cannot be scrolled in horizontal direction should not be allowed to scroll vertically. 4. We need to clip the items positioned above or left of the scrollable area. Are they correct. Please give your suggestions. Regards Madhu
Madhu M
Comment 16 Tuesday, April 3, 2007 1:51:37 AM UTC
Created attachment 13930 [details] patch for this issue patch containing fix for the overflow:hidden issue in WebKit
Madhu M
Comment 17 Tuesday, April 3, 2007 1:52:55 AM UTC
Comment on attachment 13930 [details] patch for this issue I have made the changes according to the comments given by Dave Hyatt. I have made the check for overflow clip while calculating rightmost and leftmost positions. In Firefox it is allowing the vertical scroll even if the overflow is set to hidden. So there is no need to check for the lowermost position. Now the behavior in Safari is made similar to that in Firefox. I have added two new test cases for this. Please review my patch and give the comments. Regards Madhu
Darin Adler
Comment 18 Thursday, April 5, 2007 5:17:10 PM UTC
Comment on attachment 13930 [details] patch for this issue The change logs in this patch are not good. There are double entries in each change log, and it looks like the output of the prepare-ChangeLog without any editing. That script is supposed to start a change log for you, but then you have to add the appropriate comments to explain what you're changing and why.
Dave Hyatt
Comment 19 Thursday, April 5, 2007 7:44:29 PM UTC
Comment on attachment 13930 [details] patch for this issue This is not correct. It will break Web sites. As I said before, the real fix is pretty complicated, since you have to detect that an object is entirely above or to the left of the scrollable area in order to exclude it.
Dave Hyatt
Comment 20 Thursday, May 10, 2007 10:00:43 AM UTC
Created attachment 14454 [details] Patch to fix this bug. Will land test case as layout test.
Oliver Hunt
Comment 21 Thursday, May 10, 2007 10:03:54 AM UTC
Comment on attachment 14454 [details] Patch to fix this bug. ChangeLog, and testcase if possible
mitz
Comment 22 Thursday, May 10, 2007 10:06:39 AM UTC
+ // If a positioned object lies completely to the left of our object it will be unreachable via scrolling. + // Therefore we should not allow it to contribute to the lowest position. Is this correct for RTL overflows?
Dave Hyatt
Comment 23 Thursday, May 10, 2007 10:15:02 AM UTC
Fixed in r21358
mitz
Comment 24 Thursday, May 10, 2007 10:26:08 AM UTC
Created attachment 14455 [details] RTL case broken after the fix RTL should have a vertical scroll bar too.
David Kilzer (:ddkilzer)
Comment 25 Friday, May 11, 2007 12:17:08 AM UTC
(In reply to comment #24) > Created an attachment (id=14455) [edit] > RTL case broken after the fix > > RTL should have a vertical scroll bar too. Fixed in r21360 with layout test in r21361.
Note You need to log in before you can comment on or make changes to this bug.