42187 - colspan overrides width on TD

Status: RESOLVED FIXED

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

Description

[mgsp]

From Bugzilla Helper:
User-Agent: Mozilla/4.72 [en] (X11; I; Linux 2.2.14-5.0 i586; Nav)
BuildID:    2000041805

In the page at http://www.longestjourney.com, image with girl's right hand
placed not where it should have been. It look separated from the rest of the
body.

Reproducible: Always
Steps to Reproduce:
1.go to http://www.longestjourney.com
2.not much else
3.at all

Actual Results:  incorrectly rendered page

Expected Results:  correctly rendered page

table rendering error?

Comment 1

[Blake Ross]

image looks absolutely fine to me in 61220 (i use the term 'fine' loosely); you 
were using an old build, please try again with a newer one.

Comment 2

[Joseph Elwell]

This page looks ok, as long as your Browser width is set low enough that the
girl's hand is forced into being right next to her body. However if you increase
your browser width then the hand detaches and a gap is created. That gap does
not occur in Netscape 4.X at any width.

tested using linux 2000-09-26-09-MN6 mozilla build.

Comment 3

[Asa Dotzler [:asa]]

confirming bug with 092608 mozilla trunk build on NT.  images in a table
notlining up right.  Over to HTML Tables.

Comment 4

[Scott Brodmerkle]

If you reduce the length of the comments in the Production Updates column, the
contents of the page line up much better, although still not perfectly. Might
this be related to bug 19961?

Comment 5

[mark mccormick]

duplicate bug 55850 has a nicer testcase

Comment 6

[Mats Palmgren (inactive)]

Attachment: Testcase

Comment 7

[Mats Palmgren (inactive)]

For a cell such as <td colspan="2" width="195"> the explicit width is ignored
making the cells resize when the window is resized.

Comment 8

[Bernd]

I am looking into it.

Comment 9

[Bernd]

This is really a pure table bug:
inside

PRBool 
BasicTableLayoutStrategy::ComputeNonPctColspanWidths(PRInt32           aWidthIndex,
                                                     nsTableCellFrame* aCellFrame,
                                                     nscoord           aCellWidth,
                                                     PRInt32           aColIndex,
                                                     PRInt32           aColSpan,
                                                     PRInt32&          aLimitType,
                                                     float            
aPixelToTwips)


the cell inside a fixed colspan requires more width and the desired content
width of the colspan is taken, instead of fullfilling the requirements of the
fixed cell with the minimum possible width. 

Comment 10

[Bernd]

Attachment: patch

Comment 11

[Bernd]

The attached patch fixes the testcase and the original URL. It passed the table
regression tests in core and viewertest directories. If in a near future this
'spot the loney' or 'hork the viewer' game ends, I may be able to test it even
in the remaining directories.

Comment 12

[Bernd]

the patch does not pass the regression tests in the table/bugs directory,
because it makes the regression in bug 2684 even worse. :-((

Comment 13

[karnaze (gone)]

Reassigning to bernd.

Comment 14

[Bernd]

Attachment: revised patch, fixes also bug 2684

Comment 15

[karnaze (gone)]

r=karnaze, moving to m0.9. 

Bernd, I have just a few minor whitespace style nits (in keeping with the 
style of the existing code). Could you please do

else {        
  if ((newColAdjWidth > 0) && (FIX == aWidthIndex)) {

instead of

else
{        
  if((newColAdjWidth > 0) && (FIX == aWidthIndex))
  {

Comment 16

[Marc Attinasi]

sr=[e-mail] (assuming karnaze's comments make it in).

Comment 17

[Bernd]

fix checked in

Comment 18

[Amarendra Hanumanula]

QA contact update