63916 - Tables shrink wrap alt text too tightly

Status: RESOLVED WORKSFORME

(Core :: Layout: Block and Inline, defect, P3)

Description

[Ed Leone]

From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows 98)
BuildID:    2001108

Browser:   Netscape 6.0		
Role: 			tester
Organization:	all
Course:		all
Pages/Area: 		see screen shot below
Date and Time: 	see email time stamp
Error:			When creating a new "Test/Quiz," partial text appears 
instead of the "Help" icon.  HTML code follows screen shot.

Reproducible: Always
Steps to Reproduce:
1.see description above
2.
3.

Actual Results:  see description and screen shot

Expected Results:  entire window should display

see attachments

Comment 1

[Ed Leone]

Attachment: screen shot of bug

Comment 2

[Ed Leone]

Attachment: HTML code for page

Comment 3

[timeless]

  remote = window.open(u, n, 'width=' + w + ',height=' + h 
+',resizable=yes,scrollbars=yes');
Shouldn't we show toolbars and menus and stuff?

Comment 4

[Doron Rosenberg (IBM)]

reporter, this is not a place for ns6 bugs. please get a new build and test this.

Comment 5

[Boris Zbarsky [:bzbarsky]]

I think I see the problem with a new build (2001-01-05-08 on linux).  I think I
also see the cause of the problem.  The text in question ("New Item Help") is
alt text for an image which has width="14".  The text is shown because the image
is not found.  So the text is cut off, because it's only given 14px horizontal
space.  NS 4.x does not display the text at all and puts it in a tooltip if
someone mouses over where the image should be.

Comment 6

[Boris Zbarsky [:bzbarsky]]

Over to layout for further triage.

Comment 7

[Asa Dotzler [:asa]]

reassigning. Setting bug status to New

Comment 8

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

Attachment: The real HTML, instead of an MS-HTML file containing the HTML as text within it.

Comment 9

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

Ian - is this a dupe?

Comment 10

[Hixie (not reading bugmail)]

Attachment: Simple and self explanatory testcase.

Comment 11

[Hixie (not reading bugmail)]

The bug is that if an image is found to be broken and thus resizes its frame,
tables do not reflow to take this into account. This is probably a dup of one
of the many table reflow bugs, but I couldn't find the exact one.

Comment 12

[karnaze (gone)]

In attachment #4 [details] [diff] [review], during the incremental reflow the block is not setting the 
maximum width large enough to cover "A B C" (it is not even large enough to 
cover the max element width). Reassigning to attinasi and marking m0.9.1 Tables 
rely on blocks reporting accurate maximum widths.

Tbl 00E1B07C r=1 a=11505,UC c=0,0 cnt=2228
 Tbl 00E1B0C8 r=1 a=11505,UC c=UC,UC cnt=2229
  RowG 00E1B130 r=1 a=45,UC c=45,UC cnt=2230
   Row 00E1B16C r=1 a=45,UC c=45,UC cnt=2231
    Cell 00E1B1B4 r=1 a=45,UC c=15,UC cnt=2232
     Block 00E1B210 r=1 a=15,UC c=15,UC cnt=2233
     Block 00E1B210 d=165,855 me=165 m=15
    Cell 00E1B1B4 d=195,885 me=195 m=45
   Row 00E1B16C d=45,885
  RowG 00E1B130 d=45,885
Balanced min=0 des=45 pref=45 cols=45
  RowG 00E1B130 r=2 a=45,UC c=45,UC cnt=2234
   Row 00E1B16C r=2 a=45,UC c=45,UC cnt=2235
   Row 00E1B16C d=45,885
  RowG 00E1B130 d=45,885
 Tbl 00E1B0C8 d=45,930
Tbl 00E1B07C d=525,1170

Comment 13

[Amarendra Hanumanula]

QA contact update

Comment 14

[Marc Attinasi]

Accepting

Comment 15

[karnaze (gone)]

Moving to m0.9.2

Comment 16

[Marc Attinasi]

By mandate, moving non-crashers and non-dataloss bugs out

Comment 17

[Mats Palmgren (inactive)]

Attachment: Testcase

Comment 18

[Mats Palmgren (inactive)]

Ian's testcase, attachment 22593 [details], seems to be working now so I made a new one
exposing the bug. Build 2001-08-06-03 on Windows 98 SE.

Comment 19

[Mats Palmgren (inactive)]

Is this bug perhaps related to bug 55917?

Comment 20

[Hixie (not reading bugmail)]

This might actually be a dup of the "when images pop in or break they don't 
reflow enough of the document" bug.

Comment 21

[Hixie (not reading bugmail)]

Making this bug block bug 41924, the alt text bug.

Comment 22

[Greg K.]

Reconfirmed using FizzillaCFM/2002070913. The examples in attachment 4 [details] [diff] [review] don't
look the same. Setting All/All.

Comment 23

[Bernd]

attinasi will probably not work on those bugs :-(

Comment 24

[Bernd]

This is block issue the block returns a to small maxWidth, to be correct it
returns 0. It should return 480 which appeared previously
...
          text 0236B910 d=480,228 me=132
         area 0236B8BC d=480,240 me=132 m=480
....

         text 02368C44 d=132,228 me=132
        area 0236B8BC d=132,720 me=132
       block 02368ACC d=132,720 me=132 m=0 <<<<<

Comment 25

[Martijn Wargers (dead)]

I can see the bug in a 2006-09-05 1.8 seamonkey build, but not anymore in current trunk build or a trunk build from 2006-08-01, when testing with Hixie's testcase.

Mats, I'm not sure what your testcase should be showing, can this bug be marked WORKSFORME?

Comment 26

[Mats Palmgren (inactive)]

Comment on attachment 45018 [details]
Testcase

FWIW, I can reproduce the problem this testcase shows in:
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.5+) Gecko/20011011

Comment 27

[Mats Palmgren (inactive)]

Attachment: Testcase (Standards mode)

Here's the same testcase as attachment 45018 [details] but in Standards mode.
It shows that the Alt text overflows. The bug in older builds was that it
caused the *table cell* to overflow the table boundary.

Comment 28

[Mats Palmgren (inactive)]

I think the current layout for the testcases is correct.
(Although the last one differs from what Opera and Safari do (IE untested)).

-> WORKSFORME