325486 - Input element default value invisible when overflowing parent div

Status: VERIFIED FIXED

(Core :: Layout: Form Controls, defect)

Description

[Jared McAteer]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.8) Gecko/20051111 Firefox/1.5
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.9a1) Gecko/20060201 Firefox/1.6a1

If you place an input within an absolutly positioned div that is inside another positioned div (absolute or relative, does not matter) and the input element overflows the parent div, it's default value will get cut off until the user inputs some text.

Reproducible: Always

Steps to Reproduce:
1. See test case

Actual Results:  
Input text gets cut off

Expected Results:  
Input text should show

Comment 1

[Jared McAteer]

Having no luck uploading test case, added it to url section.

Comment 2

[Ria Klaassen (not reading all bugmail)]

This is a regression between 28-Apr-2005.06 and 29-Apr-2005.06.

Comment 3

[Martijn Wargers (dead)]

I guess this could be a regression from bug 240276.

Comment 4

[Martijn Wargers (dead)]

Attachment: testcase

Attaching testcase to the bug, so no chance of losing it.

Comment 5

[philippe (part-time)]

Confirmed on OS X.

Comment 6

[chris hofmann]

roc, any chance for looking at this for 1.9a1?

Comment 7

[Robert O'Callahan (:roc) (email my personal email if necessary)]

yeah.

Comment 8

[Mike Schroepfer]

Since this looks to be a 1.8.1 regression - Roc can we get a fix in that branch?

Comment 9

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

Minusing for 1.8.1 blocking list, although we'd certainly still consider a patch.

Comment 10

[Robert O'Callahan (:roc) (email my personal email if necessary)]

The problem appears to be that around here
http://lxr.mozilla.org/mozilla/source/layout/generic/nsHTMLReflowState.cpp#2027
we set mComputedMaxWidth to something too small for the reflowState that BoxReflow is going to pass to the scrollframe0. We should not be doing this. I need to do more analysis to figure out what's wrong.

Comment 11

[Erik Fabert]

*** Bug 346192 has been marked as a duplicate of this bug. ***

Comment 12

[Robert O'Callahan (:roc) (email my personal email if necessary)]

According to the reflow state chain, the scrollframe's containing block is the innermost absolutely-positioned block. That innermost abs-pos block's reflow state has mComputedMaxWidth of 660, set by Mats' code here:
http://lxr.mozilla.org/mozilla/source/layout/generic/nsAbsoluteContainingBlock.cpp#534
The scrollframe tries to honour this mComputedMaxWidth, which causes this problem.

I'm not really sure what Mats' code is doing and it looks pretty fragile. I don't know why we are messing with the maximum width this way.

A slightly less risky option would be to hack in around here:
http://lxr.mozilla.org/mozilla/source/layout/generic/nsHTMLReflowState.cpp#2002
to go for intrinsic width for inline-block-like situations, of which this is one. I'll try that.

Comment 13

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: fix

This fixes it. The idea is to treat block children of replaced elements (where the replaced element is inside the block child's containing block) as replaced, for the purposes of setting auto-width and height, i.e., force them to use intrinsic widths and heights.

Comment 14

[Boris Zbarsky [:bzbarsky]]

The reflow branch fixes this, because we just compute the right intrinsic widths here.

Comment 15

[Peter van der Woude [:Peter6]]

fixed after reflow branch landing

Comment 16

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

Comment on attachment 232661 [details] [diff] [review]
fix

minusing this patch since it's no longer needed now that the reflow branch has landed

Comment 17

[Ryan VanderMeulen [:RyanVM]]

Adding in-testsuite? nomination per bz's request in m.d.t.l. Sorry for the bugspam.

Comment 19

[Martijn Wargers (dead)]

Attachment: reftest

The only difference between the test and the reference rendering is the width: 50px;/width: 500px;.

Comment 20

[Martijn Wargers (dead)]

Reftest checked in.
http://bonsai.mozilla.org/cvsview2.cgi?diff_mode=context&whitespace_mode=show&file=reftest.list&branch=&root=/cvsroot&subdir=mozilla/layout/reftests/bugs&command=DIFF_FRAMESET&rev1=1.21&rev2=1.22