807934 - background image not invalidating on browser resize

Status: VERIFIED FIXED

(Core :: Layout, defect)

Description

[Dennis Jakobsen]

User Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:19.0) Gecko/19.0 Firefox/19.0
Build ID: 20121101180809

Steps to reproduce:

go here: http://www.computercity.dk/default.aspx
resize browser horizontally (either way)


Actual results:

Background image not being invalidated, so it stays where it was initially rendered


Expected results:

Background image should invalidate and display correctly

Comment 1

[Loic]

Not a proper DLBI bug, it's more recent. Surely a regression in a fix about DLBI bug.

m-c
good=2012-10-26
bad=2012-10-27
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=5ecff3e46ed5&tochange=f9acc2e4d4e3

Comment 2

[Alice0775 White]

Attachment: html

Comment 3

[Alice0775 White]

Regression window(m-i)
Good:
http://hg.mozilla.org/integration/mozilla-inbound/rev/71e97a4400ea
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:19.0) Gecko/19.0 Firefox/19.0 ID:20121025182629
Bad:
http://hg.mozilla.org/integration/mozilla-inbound/rev/8a29f06470f7
Mozilla/5.0 (Windows NT 6.1; WOW64; rv:19.0) Gecko/19.0 Firefox/19.0 ID:20121025184029
Pushlog:
http://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=71e97a4400ea&tochange=8a29f06470f7

Suspected: Bug 801488

Comment 4

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

Must be the patch that removed Redraw() from nsStackLayout.

Comment 5

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

Attachment: slightly smaller testcase

Try resizing the window with this testcase open.

I'm pretty sure this dates back to DLBI and the patch in bug 801488 only exposed it because we no-longer force-redraw the entire window on resize.

Comment 6

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

http://dxr.mozilla.org/mozilla-central/layout/base/nsDisplayList.cpp.html#l2095
    // This is incorrect, We definitely need to check more things here. 
No kidding!

Comment 7

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

It seems to me that the right way to fix this might be to make nsDisplayBackgroundGeometry store (instead of mPadding/mContent) explicit background positioning area and background painting area. What do you think Matt? I need to sleep on it :-).

Comment 8

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

http://www.w3.org/TR/css3-background/#background-clip
http://www.w3.org/TR/css3-background/#the-background-origin

Comment 9

[Matt Woodrow (:mattwoodrow)]

I think we just want bug 798964 to be finished, which should implicitly includes those as the bounds of the display items. Was stuck on some bizarre android-specific failures though.

Comment 7 would probably work too though, I think I guessed with using mPadding/mContent.

Comment 10

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

(In reply to Matt Woodrow (:mattwoodrow) from comment #9)
> I think we just want bug 798964 to be finished, which should implicitly
> includes those as the bounds of the display items.

It would fix this bug, yes, but somewhat by accident, since the display item bounds would be the image bounds and the image bounds would move. I think we could have other situations where the background painting area doesn't change but the background positioning area does change, so bug 798964 wouldn't necessarily invalidate in that case. Also converting to painting area/positioning area would be more elegant.

Also, we need to fix this bug for FF18 release. Are you planning to fix bug 798964 for that?

Comment 11

[Matt Woodrow (:mattwoodrow)]

Fair call, no I don't think we should be uplifting bug 798964 to aurora.

Comment 12

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

I ended up going on a bit of a rampage. I'll probably need to come up with a different fix for Aurora...
https://tbpl.mozilla.org/?tree=Try&rev=38ac9803cf90

Comment 13

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

The reftests in that push are green.

Comment 14

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

Attachment: Simple fix

This is the really simple fix for this testcase. It should be good enough.

Comment 15

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

I'm moving my major changes to bug 810470.

Comment 16

[Matt Woodrow (:mattwoodrow)]

Fixed by bug 810470.

Comment 17

[Mihai Morar, (:MihaiMorar)]

Verified Fixed for FF 18.b4 Mozilla/5.0 (Windows NT 6.1; WOW64; rv:18.0) Gecko/20100101 Firefox/18.0 (20121212073002)

Resizing browser works as expected.

Comment 18

[Mihai Morar, (:MihaiMorar)]

Verified Fixed on FF 19.b1 on Windows 7x64 too.

Mozilla/5.0 (Windows NT 6.1; WOW64; rv:19.0) Gecko/20100101 Firefox/19.0(20130109111322)