Closed Bug 54565 Opened 24 years ago Closed 23 years ago

[QUIRKS] max element size of a block with two back-to-back images needs to be the sum of the two images in Quirks mode

Categories

(Core :: Layout, defect, P2)

defect

Tracking

()

VERIFIED FIXED
mozilla0.9.4

People

(Reporter: bugzilla, Assigned: attinasi)

References

()

Details

(Keywords: compat, testcase, topembed)

Attachments

(4 files)

Try loading the page:
http://narod.yandex.ru/404.xhtml
in IE and in Mozilla
The upper graphics in the table seems to be align to the left instead of to the 
right...
Reassigning to myself.
Assignee: clayton → karnaze
Attached file Testcase
Keywords: testcase
Another good testcase is in bug 64770 (or
http://bugzilla.mozilla.org/showattachment.cgi?attach_id=22220). Just submit the
form and look at the resulting table on Linux and then on
Windows..compare..Rinse and repeat.
Severity: normal → major
Moving to m.9.1 and moving to buster's list.
Assignee: karnaze → buster
Target Milestone: --- → mozilla0.9.1
Buster is slacking even more than usual these days, so I'm taking this bug.
Assignee: buster → attinasi
The only problem I am seeing is that the little image of the person on the left 
near the top is behind the yellow link-bar instead of in front of it. Otherwise 
the layout looks like IE and Nav. The testcase, on the other hand, is very 
different, but I fail to see what it is about - can someone please clarify?

Also, this is not a major problem, so moving to 0.9.2
Severity: major → normal
Status: NEW → ASSIGNED
Target Milestone: mozilla0.9.1 → mozilla0.9.2
This appears to have been fixed. The URL and testcase look the same as on IE in
the 0530 build. Marking WORKSFORME
Status: ASSIGNED → RESOLVED
Closed: 23 years ago
Resolution: --- → WORKSFORME
on build 20010532 on Win2k the 
http://bugzilla.mozilla.org/showattachment.cgi?attach_id=15938
doesn't look anything like the rendering in IE...:(
Status: RESOLVED → REOPENED
Resolution: WORKSFORME → ---
Looks the same for me on the Mac (using MacIE5.5). I'll try on Win2K...
OK, I tested on Win2K against Nav4.7 (IE wouldn't load it for some reason) and
it looks different, as you say. It is not too strange that MacIE gets it like
Mozilla does, since the mozilla and MacIE layout is correct. Since the second
cell is set to width=100%, the size of the first cell is decreased to it's
smallest size, and since it has two images in it, they are forced to wrap.

This is, then, a compatibility bug, since Nav treats the two images in the first
cell as unbreakable (eg. it puts them on the same line and makes the size of the
first cell big enough to fit them both). In other words, the maximum element
size of the block in the first cell is computed as the sum of the width of the
two images (instead of the correct computation which is the widest of the two
images).

Karnaze thinks that this is a dup of a very old bug, but I could not find it. 

Moving to moz1.0  Thanks Henrik!

Status: REOPENED → ASSIGNED
Keywords: compat
OS: Windows 2000 → All
Hardware: PC → All
Summary: table is rendered wrong → [QUIRKS] max element size of a block with two back-to-back images needs to be the sum of the two images in Quirks mode
Target Milestone: mozilla0.9.2 → mozilla1.0
*** Bug 52582 has been marked as a duplicate of this bug. ***
FYI, IE 5.0 on Windows 98 behaves as Nav 4.7, showing the images on one row.
*** Bug 60562 has been marked as a duplicate of this bug. ***
One testcase that I can reproduce on all platforms, which has been problematic
since 0.8,  is:

http://www.equalfooting.com

(compare with IE or 4.x)

I've opened a bug on this before, but it was closed as a dup of Bug 60562
(which has now been closed as a dup of this)
*** Bug 52699 has been marked as a duplicate of this bug. ***
Moving the priority up
Priority: P3 → P2
Target Milestone: mozilla1.0 → mozilla0.9.4
I'd love to find all of the other dups of this - there must be a lot. I have an
idea on how to fix it, at least to make it work the way IE does it...
*** Bug 74890 has been marked as a duplicate of this bug. ***
I have a fix for this now. Attaching patch shortly. Essentially, we just detect
when a line has contiguous images and accumulate the max element width for all
of the contiguous the images.
Wow! it would be nice to get this one fixed!  Too bad it didn't make it to 0.9.2
and NS6.1.
This is some quirky stuff. IE and Nav don't agree on too many of the cases I
have been testing, but in the few cases where they do agree Mozilla now is the
same. Specifically, this is the last two cases in the testcase with ID 44280 and
the testcase with ID 44278.

One notable difference between Mozilla (with the patch) and Nav/IE is the case
where the images are all in anchors, eithere all in a single anchor or all in
individual anchors:

<a href="foo"><img src="..."></a><a href="foo"><img src="..."></a>

and

<a href="foo"><img src="..."><img src="..."></a>

Mozilla (with the patch) will keep these all on the same line, whereas Nav and
IE will break them up when the table cell is specified too small. I'm twekaing
around to see if I can figure out how to emulate it, but I'm not sure how
important it is.
I just ran through all of the testcases and URLs from this bug and all of the
associated dups - HUGE improvement. Looking for some reviews now - thanks.
There are some other bugs where text that is contiguous with an image is
supposed to be kept on the same line as well (cf. bug 32191). This sounds like a
similar problem. Can we fix both at the same time? (I'd started to fix that
problem by allowing a text run to continue through an image, but at the time
wasn't clever enough to know how to update the MES and stuff.)
Hmm. I guess that a more general approach would be to look for whitespace in the 
line instead of contiguous images, and accumulate the MEW up to the whitespace? 
I'll give it a go, sounds similar to what I was trying already. I must admit 
though, it seems different than what I thought was suppossed to be happening, 
and from the HTML quotes in bug 32191 it sounds like it is not quirks mode 
behavior either...
The patch (id=44281) passes the block and regression tests. Looks like I need to
add some more tests to catch what this fixed!
More data on the regression tests: some of the table regression tests were fixed
by this patch. 

bug12910-2.html, bug14323.html, bug15544.html, and bug20804.html all had the
iamge problem and are all now correct with respect to IE and Nav. I am curious
why they were not passing before - presumably the test was for something else
and image problem was just there for the ride, but bug15544.html, for example,
looks like a perfect testcase for this bug and following the dup-trail leads me
to bug 32191 which is not yet fixed...

Alright, anyway, I'm looking at getting this patch to fix bug 32191 too, but as
it is it fixes a large class of bugs and passes the regression tests.
*** Bug 91563 has been marked as a duplicate of this bug. ***
Keywords: topembed
Marc, please look at bug 93992 too.
The patch in bug 32191 supercedes the patch attached here. The problem is more
general than just adjacent images...
Fixed along with bug 32191
Status: ASSIGNED → RESOLVED
Closed: 23 years ago23 years ago
Resolution: --- → FIXED
Marking verified in the Sept 06th build (2001-09-06-05).
Status: RESOLVED → VERIFIED
You need to log in before you can comment on or make changes to this bug.