310553 - {inc}dynamically changing font-size in fixed width absolutely positioned box causes first text node not to wrap

Status: RESOLVED WORKSFORME

(Core :: Layout: Positioned, defect)

Description

[Yuh-Ruey Chen]

In an absolutely positioned box with fixed width, the first text node or text
nodes separated only with <br>'s will not wrap after changing font-size. 

Resizing the window or min/maximizing won't fix it, but a manual "reflow" by
changing the box's width to a small value and back will.

However, this "reflow" sometimes doesn't clear the old display, leading to a
visual artifact. In fact, repositioning or resizing the box via scripting will
usually leads to such visual artifacts. The artifacts can be eliminated by
resizing or min/maximizing the window.

Aside: I'm finding so many things buggy with dynamically changing font-size. So
far I've found a couple of issues that are probably the consequences of multiple
bugs such as bug 310451, bug 271776, this, and probably more.

All the bugs are triggered by dynamically changing font-size of an absolutely
positioned box. All can be worked around by wrapping that box inside a
relatively positioned box.

Comment 1

[Yuh-Ruey Chen]

Attachment: testcase

Comment 2

[Yuh-Ruey Chen]

I should also mention that the second node (in this case, an <i> element) can
overlap with first text node. However if the first word of the second node is
large enough it will go to next line, leading to no overlap. I suspect this has
something to do with bug 271776.

Comment 3

[Yuh-Ruey Chen]

Attachment: testcase showing that only first text node is affected

Comment 4

[Yuh-Ruey Chen]

Attachment: testcase showing that only first text node is affected

Oops, last testcase was misleading. Replacement

Comment 5

[Yuh-Ruey Chen]

Attachment: testcase showing how <br> doesn't count

This testcase shows that "first text node" includes text nodes past <br>'s.

Also, interestingly, the reflow hack didn't work on the previous testcase. It
works on this one w/o leaving any visual artifacts.

Comment 6

[Mats Palmgren (inactive)]

Bug still occurs in both SeaMonkey 2006120701 (pre reflow branch) and 2006120801 (post reflow branch) on Linux, although they are broken in slightly different
ways.

Comment 7

[Yuh-Ruey Chen]

WFM as of Gecko/20090824 Firefox/3.5.3.