701656 - transform 3D don't work in transform 3D element

Status: RESOLVED FIXED

(Core :: Web Painting, defect)

Description

[Zéfling]

User Agent: Mozilla/5.0 (Windows NT 5.2; WOW64; rv:8.0) Gecko/20100101 Firefox/8.0
Build ID: 20111104165243

Steps to reproduce:

The webkit demonstration with transform-style: preserve-3d + transform: rotateY  and children with transform: translateZ don't work correctly. 3D effects is not apply on children.

http://ikilote.net/Programmation/CSS/Test/transform-style.htm

Comment 1

[Boris Zbarsky [:bzbarsky]]

Not a CSSOM issue...

Matt, could you take a look at this?

Comment 2

[Daniel Holbert [:dholbert]]

(FWIW, I get invalidation artifacts on this bug's testcase -- I filed bug 702149 on that.)

Comment 3

[Matt Woodrow (:mattwoodrow)]

Attachment: Make WrapPreserve3DList capable of walking through child display items for the same frame

Comment 4

[Matt Woodrow (:mattwoodrow)]

Attachment: Include the preserve-3d parent content in preserve-3d sorting

Comment 5

[Matt Woodrow (:mattwoodrow)]

These patches fix the majority of the problems with this test case (including the visible effects of bug 702149), though it appears we are still calculating incorrect visible regions and are clipping the transformed objects at times.

We also appear to have a different interpretation of group opacity for preserve-3d objects here, I'll put this in an email to www-style to discuss this.

Comment 6

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

Need reftests...

Comment 7

[Matt Woodrow (:mattwoodrow)]

Attachment: Add tests

Comment 8

[Matt Woodrow (:mattwoodrow)]

https://hg.mozilla.org/integration/mozilla-inbound/rev/75869e1354d0
https://hg.mozilla.org/integration/mozilla-inbound/rev/17c9363d801a
https://hg.mozilla.org/integration/mozilla-inbound/rev/bb6ef2d824a8

Comment 9

[Marco Bonardo [:mak]]

https://hg.mozilla.org/mozilla-central/rev/75869e1354d0
https://hg.mozilla.org/mozilla-central/rev/17c9363d801a
https://hg.mozilla.org/mozilla-central/rev/bb6ef2d824a8

Comment 10

[Zéfling]

Attachment: Gecko (Nightly) & Webkit (Iron) render

It's really resolved ? The result is strange in Gecko...

Comment 11

[Ryan VanderMeulen [:RyanVM]]

What release channel is your nightly build from? This was fixed only on mozilla-central (what will become Firefox 11). This will not work on nightlies from the Aurora/Beta/Release channels.

Comment 12

[Zéfling]

I'm user agent is :
Mozilla/5.0 (Windows NT 5.2; WOW64; rv:11.0a1) Gecko/20111118 Firefox/11.0a1

Is it not the last ?

Comment 13

[Ryan VanderMeulen [:RyanVM]]

Yes, that's the correct build to be testing. Thanks for confirming.

Comment 14

[Alex Keybl [:akeybl]]

[Triage Comment]
This would not block shipping FF10, but we'd consider taking a low risk patch for Aurora if nominated.

Comment 15

[Matt Woodrow (:mattwoodrow)]

Can you post the contents of your about:support please?

Comment 16

[Zéfling]

Attachment: about:support

Comment 17

[Zéfling]

Attachment: With new profile

Oh! With a new profile it's better, but the render is cut.

Comment 18

[Matt Woodrow (:mattwoodrow)]

Filed bug 707563 for the remaining Layers bug here.

Comment 19

[Zéfling]

Attachment: Gecko (Nightly) & Webkit (Iron) render

With bug 707563, the result is better, but the text is distorted. Opacity is strange, as if it were applied at all evenly.

Comment 20

[Zéfling]

Is this bug does not okay ? For the opacity, it seems that it is not a bug. I think this bug may be closed. The opacity is same for all element in flat mode and perserve-3D. It's not specify in the daft of the W3C. 

Can I close this bug ?

Comment 21

[Alex Keybl [:akeybl]]

(In reply to Célian VEYSSIÈRE from comment #20)
> Can I close this bug ?

Let's see if Matt agrees before closing this bug. Thanks.