88992 - Content ending up inside OBJECT that shouldn't be

Status: VERIFIED FIXED

(Core :: DOM: HTML Parser, defect, P1)

Description

[Ari]

The most recent releases of Mozilla (0.9 - 0.9.2 and possibly earlier) do not
display the full page and stop after drawing the image map test. Previous
versions of Mozilla were able to display the whole page. The full source of the
page is visible through the 'View Source' command.

(Yes, I know there are many bugs reports based on this one page, but it will be
a happy day indeed when Mozilla can display it perfectly.)

Comment 1

[Chris Petersen]

Confirming issue in the July 10th build (20010710).

Comment 2

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

Attachment: simpler testcase

Comment 3

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

The content model is incorrect.  "This text doesn't show up!" in the testcase
ends up inside the object element in the content model.  Changing to parser
since this seems like a content sink problem.

Comment 4

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

The content tree I'm seeing is:

html@02C0D6A0 refcount=8<
  head@02C0D798 refcount=2<
    Text@02CA4978 refcount=2<\n    >
    meta@02B8F138 http-equiv=Content-Type content=text/html; CHARSET=iso-8859-1
refcount=2<>
    title@02B67088 refcount=2<
      Text@02B670E0 refcount=2<iCab - Browser-Test>
    >
  >
  Text@02B8EEA8 refcount=3<\n>
  body@02B68290 refcount=3<
    Text@02B77FA0 refcount=3<\n\n>
    p@02B70058 refcount=3<
      Text@02D09048 refcount=4<Embeding an image using the <OBJECT> tag. In addi
tion an image map is defined where\nall the links are automatically offered as n
ormal links if the browser is not able to embed the image.>
    >
    Text@02D09200 refcount=3<\n\n>
    object@02D092C0 data=http://www.icab.de/img/test.gif type=image/gif width=32
0 height=240 border=2 usemap=#map refcount=5<
      Text@02D09738 refcount=2<\n>
      p@02D097A0 refcount=2<
        Text@02D09860 refcount=2<Your browser doesn't support embeding images us
ing the OBJECT tag.\n>
        map@02D099E0 name=map refcount=4<
          p@02D09BE0 refcount=2<
            Text@02D09CA0 refcount=2<\n>
            a@02D09DE0 shape=rect coords=67,94,282,147 href=#blau title=Blaues R
echteck refcount=3<
              Text@02D0A0E0 refcount=2<Blaues Rechteck>
            >
            Text@02D09C38 refcount=2<\n>
          >
        >
        Text@02D097F8 refcount=2<\n\n\n>
      >
      p@02D100D8 refcount=2<
        Text@02D10130 refcount=2<This text doesn't show up.!>
      >
      Text@02D101E0 refcount=2<\n\n>
    >
  >
>

Comment 5

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

http://validator.w3.org/check?uri=http%3A%2F%2Fbugzilla.mozilla.org%2Fshowattachment.cgi%3Fattach_id%3D41982&doctype=Inline&ss=&sp=&noatt=
shows the correct content model

Comment 6

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

*** Bug 89035 has been marked as a duplicate of this bug. ***

Comment 7

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

*** Bug 51974 has been marked as a duplicate of this bug. ***

Comment 8

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

See also bug 53525 on the same page (via duplicate bug 51974).

Comment 9

[hirata masakazu]

*** Bug 91846 has been marked as a duplicate of this bug. ***

Comment 10

[hirata masakazu]

*** Bug 92004 has been marked as a duplicate of this bug. ***

Comment 11

[hirata masakazu]

Please take notice of the test cases and analyses in bug 92004.

Comment 12

[mitchejj]

Someone should change the platform and OS as well as the severity... after all
it does affect Mac OS X as well as others I would assume. I would also assume
that this bug effects other pages validated as correct webpages (w3c thing). A
page not  rendering fully is a big problem.

Comment 13

[Greg K.]

For convenience, I shall reiterate my analysis and testcases herein.

To quote from bug 92004:

"When an OBJECT element contains alternate content and that alternate content is
marked up only by an opening tag (the closing tag being omitted as optional),
all markup following the OBJECT element is not displayed.

Reproducible: Always
Steps to Reproduce:
1. Access the test URL or attached testcase 1.

Actual Results:  In both the test URL and the attached testcase, though there is
markup following the OBJECT element, it is not displayed unless the optional
closing tag is added to the alternate content inside the OBJECT.

Expected Results:  Any and all markup following the OBJECT element should have
been displayed.

Attached testcase 2 includes the optional closing tag on the alternate content
inside the OBJECT element, and demonstrates that the content following the
OBJECT element is now displayed."

Comment 14

[Greg K.]

Attachment: Testcase 1: Most minimal testcase exhibiting problem

Comment 15

[Greg K.]

Attachment: Testcase 2: testcase 1 file resolved by closing alternate markup inside OBJECT element

Comment 16

[Greg K.]

Marking all/all.

Comment 17

[Greg K.]

Adding html4 keyword as testcase that fails is valid HTML 4.0 code per W3.org
validator.

Comment 18

[Greg K.]

Marking Major as a valid HTML 4.0 page fails to display completely

Comment 19

[harishd]

Setting to m0.9.3.

Comment 20

[harishd]

Moving to more realistic target - m0.9.4!

Comment 21

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

*** Bug 92095 has been marked as a duplicate of this bug. ***

Comment 22

[harishd]

Attachment: patch v1.1 [ A safe fix...though not my favorite one :-( ]

Comment 23

[harishd]

The actual problem seems to be with the element table. Where mParentBits is used
for identifying elements' group and also for representing their parent model.
Fixing this would be risky.

FYI: I did try to fix it but my regression tests did not pass completely and
hence settled down with the safer fix.

Comment 24

[harishd]

Attachment: patch v1.2 [ Made OBJECT a block closure to handle all unclosed block elements within OBJECT ]

Comment 25

[Eric Pollmann]

r=pollmann, this fix looks good!  :)

Comment 26

[vidur (gone)]

Confirmed that the fix still allows OBJECTs to be inline elements. The gClosers
list contains elements that close elements within themselves, not before. sr=vidur.

Comment 27

[Christopher Hoess (gone)]

*** Bug 94158 has been marked as a duplicate of this bug. ***

Comment 28

[harishd]

Fix landed on 08/07/01. Marking FIXED.

Comment 29

[bsharma]

Verified on 2001-08-08-Trunk build on WinNT

The above url and the attached test cases loads fine.