1147673 - nsDisplayWrapList bounds unnecessarily large when it contains a scrollbox with a display port

Status: RESOLVED FIXED

(Core :: Layout, defect, P1)

Description

[Markus Stange [:mstange]]

If you have something like this:

<div style="opacity: 0.9">
  <div class="width:400px; height: 200px; overflow: auto;"
       reftest-displayport-x="0" reftest-displayport-y="0"
       reftest-displayport-w="2000" reftest-displayport-h="200">
    <div style="background: red; width: 1000px; height: 100px;">
    </div>
  </div>
</div>

Then the bounds for the nsDisplayOpacity include the whole 1000px of width of the scrollbox contents, even though those contents are clipped to a width of 400px.

Comment 1

[Markus Stange [:mstange]]

The reftest pull-background-displayport-5.html fails because of this.

Comment 2

[Markus Stange [:mstange]]

Fixing this will fix bug 1122916.

Comment 3

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=3bbdd998e80b

Comment 4

[Botond Ballo [:botond]]

Attachment: MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Comment 5

[Botond Ballo [:botond]]

Comment on attachment 8669195 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Markus, let me know if this patch resolves your issue. It makes fixed-pos-scrollable1.html pass for me locally (and all the async-scroll reftests, until I run into an assertion in split-opacity-layers-1.html, which looks unrelated (to this patch, possibly related to your patches)).

Comment 6

[Markus Stange [:mstange]]

Comment on attachment 8669195 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Works perfectly, thanks!

I'll look into the remaining test failures and assertions.

Comment 7

[Botond Ballo [:botond]]

Comment on attachment 8669195 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Comment 8

[Kartikaya Gupta (email:[e-mail])]

Comment on attachment 8669195 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

https://reviewboard.mozilla.org/r/21143/#review19127

Comment 9

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=2ab13dff70b9

Comment 10

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=dee05b9dccfc

Comment 11

[Markus Stange [:mstange]]

Attachment: MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r?botond

Comment 12

[Markus Stange [:mstange]]

Attachment: MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Comment 13

[Markus Stange [:mstange]]

Attachment: MozReview Request: Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r?tn

Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r?tn

Comment 14

[Markus Stange [:mstange]]

Attachment: MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r?tn, r?roc

Bug 1147673 - Make display items know about their scroll clips. r?tn, r?roc

Comment 15

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

Comment on attachment 8671860 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r?tn, r?roc

https://reviewboard.mozilla.org/r/21649/#review19501

::: layout/base/DisplayListClipState.h:19
(Diff revision 1)
> +class ScrollClipStack {

Please add a comment explaining what this class is for/what it means.

::: layout/base/DisplayListClipState.h:32
(Diff revision 1)
> +  {

Add a comment explaining what this function is for.

Comment 16

[Markus Stange [:mstange]]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Bug 1147673 - Make display items know about their scroll clips. r?tn, r?roc

Comment 17

[Markus Stange [:mstange]]

Attachment: MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r?botond

Comment 18

[Markus Stange [:mstange]]

Attachment: MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Comment 19

[Markus Stange [:mstange]]

Attachment: MozReview Request: Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r=tn

Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r?tn

Comment 20

[Markus Stange [:mstange]]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Bug 1147673 - Make display items know about their scroll clips. r?tn, r?roc

Comment 21

[Markus Stange [:mstange]]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Bug 1147673 - Make display items know about their scroll clips. r?tn, r?roc

Comment 22

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

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

https://reviewboard.mozilla.org/r/21643/#review19571

::: layout/base/DisplayItemScrollClip.h:32
(Diff revision 4)
> + * scroll frame.

Excellent!

::: layout/base/DisplayItemScrollClip.h:53
(Diff revision 4)
> +   * Both aPreferred and aOther are allowed to be null.

This isn't quite clear enough. You say it's "analogous" to the intersection operation, but it doesn't actually perform an intersection operation, so you need to say exactly what you mean by "the most restrictive one"? It seems there's an assumption here that one of aPreferred and aOther is contained by the other, so document that and assert it.

Comment 23

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=9e09540a588b

Comment 24

[Markus Stange [:mstange]]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Bug 1147673 - Make display items know about their scroll clips. r?tn, r?roc

Comment 25

[Botond Ballo [:botond]]

Comment on attachment 8672636 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

https://reviewboard.mozilla.org/r/21743/#review19751

Kats already r+'d this patch before he went on PTO. There's no need to review it again.

Comment 26

[Botond Ballo [:botond]]

(In reply to Botond Ballo [:botond] from comment #25)
> https://reviewboard.mozilla.org/r/21743/#review19751
> 
> Kats already r+'d this patch before he went on PTO.

(In comment 8.)

Comment 27

[Botond Ballo [:botond]]

Comment on attachment 8672635 [details]
MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

https://reviewboard.mozilla.org/r/21741/#review19775

This is fine for the case where AlignFixedAndStickyLayers unadjusts the layer's own clip, but what about a case where AlignFixedAndStickyLayers adjusts the clip of a descendant layer? The ApplyAsyncContentTransformToTree call for that descendant layer will have already happened, and potentially intersected its scroll clip with its not-yet-unadjusted clip.

Comment 28

[Markus Stange [:mstange]]

(In reply to Botond Ballo [:botond] from comment #27)
> This is fine for the case where AlignFixedAndStickyLayers unadjusts the
> layer's own clip, but what about a case where AlignFixedAndStickyLayers
> adjusts the clip of a descendant layer? The ApplyAsyncContentTransformToTree
> call for that descendant layer will have already happened, and potentially
> intersected its scroll clip with its not-yet-unadjusted clip.

Let's call the container layer L and the descendant D. L has a frame metrics for scroll id SL. D is a fixed layer that has a fixedness annotation with fixed-to-scrollid SL. But D does not have a framemetrics for scroll id SL, because a layer never shares a framemetrics with any of its ancestors. So we won't have applied a scroll clip from an SL framemetrics on D.
More generally, I think we can say that a layer never gets a scroll clip from a scroll id for which its ancestor will unadjust, because that would require having a frame metrics of the same scroll id on a layer and its ancestor. So the case you brought up can never occur.
Does that sound right?

Comment 29

[Markus Stange [:mstange]]

Comment on attachment 8672635 [details]
MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21741/diff/1-2/

Comment 30

[Markus Stange [:mstange]]

Comment on attachment 8672636 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21743/diff/1-2/

Comment 31

[Markus Stange [:mstange]]

Comment on attachment 8672637 [details]
MozReview Request: Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r=tn

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21745/diff/1-2/

Comment 32

[Markus Stange [:mstange]]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21643/diff/5-6/

Comment 33

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=c492cddc3f2f

Comment 34

[Botond Ballo [:botond]]

(In reply to Markus Stange [:mstange] from comment #28)
> (In reply to Botond Ballo [:botond] from comment #27)
> > This is fine for the case where AlignFixedAndStickyLayers unadjusts the
> > layer's own clip, but what about a case where AlignFixedAndStickyLayers
> > adjusts the clip of a descendant layer? The ApplyAsyncContentTransformToTree
> > call for that descendant layer will have already happened, and potentially
> > intersected its scroll clip with its not-yet-unadjusted clip.
> 
> Let's call the container layer L and the descendant D. L has a frame metrics
> for scroll id SL. D is a fixed layer that has a fixedness annotation with
> fixed-to-scrollid SL. But D does not have a framemetrics for scroll id SL,
> because a layer never shares a framemetrics with any of its ancestors. So we
> won't have applied a scroll clip from an SL framemetrics on D.
> More generally, I think we can say that a layer never gets a scroll clip
> from a scroll id for which its ancestor will unadjust, because that would
> require having a frame metrics of the same scroll id on a layer and its
> ancestor. So the case you brought up can never occur.
> Does that sound right?

I agree that D will never have a scroll clip for SL. But what if D is itself scrollable, and has its own scroll clip as a result?

Comment 35

[Botond Ballo [:botond]]

Comment on attachment 8672635 [details]
MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

Dropping review flag until we clear up comment 34. (It's very possible that I'm the one missing something - if so, my apologies.)

Comment 36

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=5fabc68a629e

Comment 37

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=5516169b74fc

Comment 38

[Markus Stange [:mstange]]

Let's say D (the fixed layer) is also scrollable, with scroll id SD. When ApplyAsyncContentTransformToTree is called for D, it will set a clip on D - the scroll clip for SD. At this point it doesn't know anything about any future adjustments for SL, so with its current model of the world, the layer position and its clip, and in particular the offset between the layer and the clip, are correct.
Now ApplyAsyncContentTransformToTree is called for L. An async scroll offset is applied to L, which causes D and its clip to move with it on the screen. Then AlignFixedAndStickyLayers comes along and unadjusts both D's position and its clip, and the offset between the clip and the layer stays constant (and we've already determined that this offset was correct). So I think the result will be correct, too.

I'll try to answer your original question from a different perspective again, mostly for myself.

(In reply to Botond Ballo [:botond] from comment #27)
> This is fine for the case where AlignFixedAndStickyLayers unadjusts the
> layer's own clip, but what about a case where AlignFixedAndStickyLayers
> adjusts the clip of a descendant layer? The ApplyAsyncContentTransformToTree
> call for that descendant layer will have already happened, and potentially
> intersected its scroll clip with its not-yet-unadjusted clip.

Both that scroll clip and the regular layer clip will have been not-yet-unadjusted at that point. Any future unadjusting by the container will be for an ancestor frame metrics, and thus needs to apply to both the regular layer clip and the already-applied scroll clip. Since it needs to happen to both, we can just unadjust the intersection.

Comment 39

[Botond Ballo [:botond]]

Thanks, Markus, for the explanation. After thinking about this some more, I agree. Your patch fixes a problem that only comes up in the first place in cases where a fixed/sticky layer is its own subtree root.

Comment 40

[Botond Ballo [:botond]]

Comment on attachment 8672635 [details]
MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

https://reviewboard.mozilla.org/r/21741/#review22505

Comment 41

[Botond Ballo [:botond]]

Comment on attachment 8672635 [details]
MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

https://reviewboard.mozilla.org/r/21741/#review22507

Actually, I'm now confused again.

In the case where a fixed/sticky layer is its own subtree root, AlignFixedAndStickyLayers doesn't adjust the layer's own clip (because 'adjustClipRect' is conditioned on 'aLayer != aTransformedSubtreeRoot'). Given that, it seems like this patch should be a no-op. If it's not, then I'm still missing something.

Comment 42

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=777ebc3a6dc3

Comment 43

[Botond Ballo [:botond]]

(In reply to Botond Ballo [:botond] from comment #41)
> Actually, I'm now confused again.
> 
> In the case where a fixed/sticky layer is its own subtree root,
> AlignFixedAndStickyLayers doesn't adjust the layer's own clip (because
> 'adjustClipRect' is conditioned on 'aLayer != aTransformedSubtreeRoot').
> Given that, it seems like this patch should be a no-op. If it's not, then
> I'm still missing something.

Ok, confusion resolved after talking over IRC!

The patch I posted to this bug, which is going to land along with Markus' patches, changes the condition from 'aLayer != aTransformedSubtreeRoot' to 'aTransformAffectsLayerClip'. This can be true even in cases where the fixed/sticky layer is its own subtree root, if the layer has a Layer::mClipRect, or if the layer has multiple FrameMetrics and a metrics below the one being transformed has a FrameMetrics::mClipRect (i.e. if asyncClip.isSome() is true at the point of the AlignFixedAndStickyLayers call).

Therefore, the patch is not a no-op, and by your reasoning in comment 38, is correct. Sorry it took so long for me to arrive at this conclusion!

Comment 44

[Botond Ballo [:botond]]

Comment on attachment 8672635 [details]
MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

https://reviewboard.mozilla.org/r/21741/#review22697

Comment 45

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=cd3aeed98e0f

Comment 46

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=c22fd01eada3

Comment 47

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=9b4da00a5b17

Comment 48

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=bc8054579178

Comment 49

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=c76ddc1ec63c

Comment 50

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=d01132d2ec72

Comment 51

[Botond Ballo [:botond]]

(Clearing old needinfo.)

Comment 52

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=51aed43bb546

Comment 53

[Markus Stange [:mstange]]

(In reply to Markus Stange [:mstange] from comment #50)
> https://treeherder.mozilla.org/#/jobs?repo=try&revision=d01132d2ec72

This one is green! I'll attach the patches for final review now. They could probably be simplified a little more but I'm afraid to break anything.

Comment 54

[Markus Stange [:mstange]]

Comment on attachment 8672635 [details]
MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21741/diff/2-3/

Comment 55

[Markus Stange [:mstange]]

Comment on attachment 8672636 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21743/diff/2-3/

Comment 56

[Markus Stange [:mstange]]

Comment on attachment 8672637 [details]
MozReview Request: Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r=tn

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21745/diff/2-3/

Comment 57

[Markus Stange [:mstange]]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21643/diff/6-7/

Comment 58

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=b70299aa04b9

Comment 59

[Markus Stange [:mstange]]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21643/diff/7-8/

Comment 60

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=ed5deb13fb1a

Comment 61

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=4c9337570a57

Comment 62

[Markus Stange [:mstange]]

Comment on attachment 8672635 [details]
MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21741/diff/3-4/

Comment 63

[Markus Stange [:mstange]]

Comment on attachment 8672636 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21743/diff/3-4/

Comment 64

[Markus Stange [:mstange]]

Comment on attachment 8672637 [details]
MozReview Request: Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r=tn

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21745/diff/3-4/

Comment 65

[Markus Stange [:mstange]]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21643/diff/8-9/

Comment 66

[Kartikaya Gupta (email:[e-mail])]

Comment on attachment 8672636 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Doesn't seem like anything changed from the interdiff.

Comment 67

[Kartikaya Gupta (email:[e-mail])]

(Inheriting P1 from bug 1122916)

Comment 68

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=e65b457986cf

Comment 69

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=e4e74136ec78

Comment 70

[Markus Stange [:mstange]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=71ae03bce233

Comment 71

[Timothy Nikkel (:tnikkel)]

Comment on attachment 8672637 [details]
MozReview Request: Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r=tn

https://reviewboard.mozilla.org/r/21745/#review24921

Comment 72

[Timothy Nikkel (:tnikkel)]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

https://reviewboard.mozilla.org/r/21643/#review24923

Comment 73

[Markus Stange [:mstange]]

https://hg.mozilla.org/integration/mozilla-inbound/rev/1cbfa21cf2b52cb5f01360bed12aeab53fd26d76
Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

https://hg.mozilla.org/integration/mozilla-inbound/rev/9e1f2694536c182e853f3acff4675ad5c07d7669
Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

https://hg.mozilla.org/integration/mozilla-inbound/rev/04b102bec36340d84ef40658fae4a0856d762540
Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r=tn

https://hg.mozilla.org/integration/mozilla-inbound/rev/68b33692bed3eb35dee9184b7b64daaf93c2ca9a
Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Comment 74

[Carsten Book [:Tomcat]]

sorry had to back this out for test failures like https://treeherder.mozilla.org/logviewer.html#?job_id=18633322&repo=mozilla-inbound

Comment 75

[Pulsebot]

Backout:
https://hg.mozilla.org/integration/mozilla-inbound/rev/7001bc09174e
https://hg.mozilla.org/integration/mozilla-inbound/rev/89edfed8aa0e
https://hg.mozilla.org/integration/mozilla-inbound/rev/2a6fc436951b
https://hg.mozilla.org/integration/mozilla-inbound/rev/0e2aa1eab41c

Comment 76

[Carsten Book [:Tomcat]]

also seems https://treeherder.mozilla.org/logviewer.html#?job_id=18632538&repo=mozilla-inbound was related to this changes here

Comment 77

[Markus Stange [:mstange]]

The mulet reftest failure was because I wasn't resetting the current scroll clip when entering an nsDisplayZoom.

Comment 78

[Markus Stange [:mstange]]

Comment on attachment 8672635 [details]
MozReview Request: Bug 1147673 - Unadjust clip before intersecting it with the scroll clip. r=botond

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21741/diff/4-5/

Comment 79

[Markus Stange [:mstange]]

Comment on attachment 8672636 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21743/diff/4-5/

Comment 80

[Markus Stange [:mstange]]

Comment on attachment 8672637 [details]
MozReview Request: Bug 1147673 - Use ancestor clip for root scrollable framemetrics clips. r=tn

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21745/diff/4-5/

Comment 81

[Markus Stange [:mstange]]

Comment on attachment 8671857 [details]
MozReview Request: Bug 1147673 - Make display items know about their scroll clips. r=tn, r=roc

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/21643/diff/9-10/

Comment 82

[Markus Stange [:mstange]]

Attachment: MozReview Request: Bug 1147673 - Relax the ancestor transform assertion a little. r?kats

Review commit: https://reviewboard.mozilla.org/r/28885/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/28885/

Comment 83

[Kartikaya Gupta (email:[e-mail])]

Comment on attachment 8672636 [details]
MozReview Request: Bug 1147673 - Determine more accurately whether an async transform affects a layer's clip rect. r=kats

https://reviewboard.mozilla.org/r/21743/#review25649

Comment 84

[Kartikaya Gupta (email:[e-mail])]

Comment on attachment 8701076 [details]
MozReview Request: Bug 1147673 - Relax the ancestor transform assertion a little. r?kats

https://reviewboard.mozilla.org/r/28885/#review25651

Comment 85

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/f8bea1c5413d
https://hg.mozilla.org/integration/mozilla-inbound/rev/939fb81b54e1
https://hg.mozilla.org/integration/mozilla-inbound/rev/db3a7c3208e7
https://hg.mozilla.org/integration/mozilla-inbound/rev/7d23d87b02f8
https://hg.mozilla.org/integration/mozilla-inbound/rev/bb5becd378f4

Comment 86

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/f8bea1c5413d
https://hg.mozilla.org/mozilla-central/rev/939fb81b54e1
https://hg.mozilla.org/mozilla-central/rev/db3a7c3208e7
https://hg.mozilla.org/mozilla-central/rev/7d23d87b02f8
https://hg.mozilla.org/mozilla-central/rev/bb5becd378f4