536061 - -moz-box-shadow doesn't well-support -moz-transform: translate() and/or rotate()

Status: RESOLVED FIXED

(Core :: Web Painting, defect, P2)

Description

[Ivan Enderlin]

User-Agent:       Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US; rv:1.9.3a1pre) Gecko/20091219 Minefield/3.7a1pre
Build Identifier: 

When applying a -moz-box-shadow on a -moz-transform: translate(), the shadow isn't well-placed. It's shifted.
Look the source: <http://mozilla.hoa-project.net/Bug/Transform_and_shadow.html> or the PNG : <http://mozilla.hoa-project.net/Bug/Transform_and_shadow.png>.

Reproducible: Always

Comment 1

[Emil Ivanov]

works: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1.6) Gecko/20091201 Firefox/3.5.6

broken:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2b6pre) Gecko/20091231 Namoroka/3.6b6pre

broken:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.3a1pre) Gecko/20091231 Minefield/3.7a1pre

Comment 2

[Olli Pettay [:smaug][[e-mail]]]

Regression range would be great.

Comment 3

[Emil Ivanov]

build range

works:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2b1pre) Gecko/20091006 Namoroka/3.6b1pre

broken:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2b1pre) Gecko/20091007 Namoroka/3.6b1pre

Comment 4

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

Bug 513082 and maybe bug 514670 seem like possibilities after reading http://hg.mozilla.org/releases/mozilla-1.9.2/pushloghtml?startdate=2009-10-06+00%3A00%3A00&enddate=2009-10-07+08%3A00%3A00

Comment 5

[Ivan Enderlin]

I've changed the bug's importance from normal to blocker (because of its regression nature).

Comment 6

[Mike Beltzner [:beltzner, not reading bugmail]]

My instinct is that this isn't a hard release blocker, and to clean it up in a 3.6.x release, but I don't know how common these sorts of things are. I would have expected to see more web-compatibility bugs if it was a common pattern, though.

David, Roc: thoughts on blocking?

Comment 7

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

It's unlikely to be a Web-compatibility problem since it's in the interaction of two features that aren't supported in IE.  But it could affect things people are working on using those new features.

It would be good to know what the underlying problem is, but I think it's probably something that we could fix in a .1 release.

Comment 9

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

inset shadows seem to be OK, so I'm guessing that maybe this is related to nsDisplayBoxShadowOuter::ComputeVisibility

Comment 10

[Michael Ventnor]

So this used to work? That's odd, since bug 514670 only changed behaviour if the frame had border radii.

Comment 11

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

I think bug 513082 is the more likely cause.

Comment 12

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

Attachment: reporter's testcase

Comment 13

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

The mozilla-central regression range, from Linux x86-64 nightlies, is:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=db6d2486737d&tochange=3756cdca8b54

Comment 14

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

I confirmed via try server that http://hg.mozilla.org/mozilla-central/rev/b540248e3163 is broken and http://hg.mozilla.org/mozilla-central/rev/6bfb2b05290f works, which means this is in fact a regression from http://hg.mozilla.org/mozilla-central/rev/b540248e3163 .

Comment 16

[dysinscr]

Attachment: box-shadow and transformation bugs

this attachment shows the -moz-box-shadow bug when using it with -moz-transform.
webkit and opera have a good behavior.
firefox nitghly x86 and x86_64 Linux builds are also affected.

Comment 17

[Daniel Kirsch]

See also testcase in duplicate bug 561119 - attachment 440798 [details]

Comment 21

[Mats Palmgren (inactive)]

Attachment: Patch rev. 1

Simple fix.

Comment 22

[gadjo]

There's only reftest for -moz-transform: translate(), shouldn't the test for -moz-transform: rotate() be included as well?

Comment 23

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

Comment on attachment 490561 [details] [diff] [review]
Patch rev. 1

Thanks!

Comment 24

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

(In reply to comment #22)
> There's only reftest for -moz-transform: translate(), shouldn't the test for
> -moz-transform: rotate() be included as well?

It doesn't really matter, the underlying code paths are the same.

Comment 25

[Mats Palmgren (inactive)]

Attachment: Patch rev. 1 + some more tests

Will push as soon as my Try run is green...

Comment 26

[Mats Palmgren (inactive)]

http://hg.mozilla.org/mozilla-central/rev/dab3886f5e70

Comment 28

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

I tried backporting the patch to 1.9.2, but it didn't seem to work.  Maybe I did something wrong, though.

And it looks like on 1.9.2 we need to fix the is-themed case in nsDisplayBackground as well.  (I see transformed form controls failing in a similar way.)