931853 - negative margin larger than the height confuses vertical alignment of the frame on the line

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[keith]

Attachment: bug.zip

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:24.0) Gecko/20100101 Firefox/24.0 (Beta/Release)
Build ID: 20130910160258

Steps to reproduce:

image = 350x250 inside a div with negative margin

see the sample html in the zip file. in firebug.. add the negative margin up to -600.
you will see that when the image is at negative 250 , it will start to go down. instead of going up.


Actual results:

as you have noticed. margin-top = -100px is has the same effect with -400px

if the margin is >= 0-250 (image height) = ok
if the margin is between -250 to -500 (twice image height) = BUG
if the margin -500 above = ok


Expected results:

if the margin is between -250 to -500 (twice image height) = should be not show the image

Comment 1

[Loic]

Attachment: bug.jpg

Comment 2

[Loic]

Attachment: testcase_931853.html

Comment 3

[Mats Palmgren (inactive)]

Clamping a negative logical height in nsLineLayout::VerticalAlignFrames seems to fix it:
https://tbpl.mozilla.org/?tree=Try&rev=334398ffa496

Comment 4

[Mats Palmgren (inactive)]

Shifting it into mAscent is even better since we want the frame to move away further
with more negative margin.
https://tbpl.mozilla.org/?tree=Try&rev=09e03af8c710

Comment 5

[Mats Palmgren (inactive)]

Attachment: fix+tests

The bug only occurs in quirks mode.  All tested UAs, Firefox, IE,
Chrome(blink), Safari(webkit) and Opera(presto) are compatible in
standards mode.

For the quirks mode testcase:

Chrome, Safari, and IE9 matches what this patch implements -- the
element moves further up with increasingly larger negative margin-top.

IE10 and and Opera matches what I implemented in comment 3, IOW
remove the "pfd->mAscent -= logicalHeight" line from this patch.

(if I force "IE5 quirks" mode in IE10 then I get the IE9 quirks mode
layout; I'm not sure if web pages can force that though)

https://tbpl.mozilla.org/?tree=Try&rev=e2cd1a487afb

Comment 6

[Mats Palmgren (inactive)]

https://tbpl.mozilla.org/?tree=Try&rev=3c9f18500605

Comment 7

[Mats Palmgren (inactive)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/475bd77c3400

Comment 8

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/475bd77c3400