41635 - Navquirks: background-repeat and background-position fail on tables

Status: VERIFIED FIXED

(Core :: Layout: Tables, defect, P3)

Description

[Ryan Hayle]

I appologize if I've assigned this to the wrong section--I was suprised not to
find one labeled "CSS" or something similar.

I have added a background image to a table via CSS like so:

<table width="100%" border="0" cellspacing="12" cellpadding="0" style=
"background-image:url('images/radar-bg.gif'); background-repeat:no-repeat;
background-position: bottom right">

This renders fine in *ahem* IE5, however in Mozilla (2000060508) the image is
repeated (tiled).  Of course, because of this it is thus not honoring the bottom
-right position property either.

Comment 1

[Hixie (not reading bugmail)]

Erm, this works fine for me:
   http://www.bath.ac.uk/%7Epy8ieh/m/fixedpos-bg-tables.html

Could you attach a test case, or give the URI of a test case, which shows the
problem? If possible, make the page qutie small (less than 2kb).

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: testcase

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

This bug is in quirks-mode only and is specific to tables (or perhaps other 
table elements - haven't tested).

Do table backgrounds in NavQuirks mode really need to be so quirky?  IMO, quirks 
mode should only be for quirks that pages depend on, not for perfect imitation 
of NN 4.x behavior.

Changing component to HTMLTables, confirming, and reassigning.

Comment 4

[karnaze (gone)]

This bug has been marked "future" because the original netscape engineer working 
on this is over-burdened. If you feel this is an error, that you or another 
known resource will be working on this bug,or if it blocks your work in some way 
-- please attach your concern to the bug for reconsideration. 

Comment 5

[Bernd Mielke]

Adding testcase keyword so this doesn't show up on the bugathon search page.

Comment 6

[Bernd]

Analysis:

The recently deceased Nav4xx does not display background images that are
attached via CSS to a table. So it was decided (definetely  not by me) to
disable these background images in the quirk's mode.

To illustrate this look at:

http://lxr.mozilla.org/seamonkey/source/layout/html/table/src/nsTableColFrame.cpp#124

http://lxr.mozilla.org/seamonkey/source/layout/html/table/src/nsTableColGroupFrame.cpp#451

http://lxr.mozilla.org/seamonkey/source/layout/html/table/src/nsTableFrame.cpp#1332
http://lxr.mozilla.org/seamonkey/source/layout/html/table/src/nsTableRowFrame.cpp#537

http://lxr.mozilla.org/seamonkey/source/layout/html/table/src/nsTableRowGroupFrame.cpp#212

The only place where the nsCSSRendering::PaintBackground is executed regardless
of the
quirk-mode is nsTableCellFrame.cpp because by this the background of the table
cell's
is drawn. The bad thing is that nsCSSRendering::PaintBackground makes two
things: it
first applies a background image if present and in the second step applies the
requested background color.  The patch tells the nsCSSRendering::PaintBackground
routine
that for this cell there is no background image.

I tested the patch against the
- precheckin tests
- all table regression test's

Solution:

Index: nsTableCellFrame.cpp
===================================================================
RCS file: /cvsroot/mozilla/layout/html/table/src/nsTableCellFrame.cpp,v
retrieving revision 3.226
diff -u -r3.226 nsTableCellFrame.cpp
--- nsTableCellFrame.cpp	2000/10/28 22:14:57	3.226
+++ nsTableCellFrame.cpp	2000/11/18 14:12:07
@@ -304,7 +304,11 @@
       const nsStyleTable* cellTableStyle;
       GetStyleData(eStyleStruct_Table, ((const nsStyleStruct *&)cellTableStyle));
       nsRect  rect(0, 0, mRect.width, mRect.height);
-
+
  nsCompatibility mode;
+      aPresContext->GetCompatibilityMode(&mode);
+      if (eCompatibility_NavQuirks == mode) {
+
	  ((nsStyleColor*)myColor)->mBackgroundImage.SetLength(0);
+
  } // Don't render background images in quirks mode
       // only non empty cells render their background
       if (PR_FALSE == GetContentEmpty()) {
         nsCSSRendering::PaintBackground(aPresContext, aRenderingContext, this,
 

Comment 7

[Bernd]

Probably my patch is more a quick shot, that could have been applied before rtm 
(relase the monkey) but a better way would be to 
split nsCSSRendering::PaintBackground into nsCSSRendering::PaintBackgroundImage 
and nsCSSRendering::PaintBackgroundColor and to call both function's from 
PaintBackground so that nothing changes for regular call's of 
nsCSSRendering::PaintBackground. By doing this one could enable the background 
images also in the quirk mode and maintain the coloring as it was before. 

I tried to run a regression test without all these quirk mode stuff for the 
background painting and so many differences occure that I would be very afraid 
to enable the full background painting also in the quirk mode.
 
 
 
 

Comment 8

[Bernd]

*** Bug 63575 has been marked as a duplicate of this bug. ***

Comment 9

[cdavison]

Attachment: screenshot of broken table background image in http://www.securityfocus.com/

Comment 10

[Amarendra Hanumanula]

QA contact update

Comment 11

[Markus Hübner]

another URL to see this one is
http://www.world-direct.com/central

Comment 12

[fantasai]

FIXED by Bernd Mielke's checkin on 2001-07-01 (bug 46268)

Comment 13

[Amarendra Hanumanula]

 Marking verified on 2001-07-27-00-0.9.2 build

Comment 14

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

That's interesting, since the checkin wasn't even on the 0.9.2 branch.

Comment 15

[Markus Hübner]

the testcase at
http://www.world-direct.com/central/
is broken with the final Netscape 6.1 

How can that be?!

Comment 16

[Bernd]

very easily, this is the consequence of checking in to the trunk only.