Make slots in non-shadow trees participate in a flat tree

Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in real world.

To make the code simple and stable, it would be better to make a slot in a
non-shadow tree participate in a flat tree, as normal elements.

Note that this new behavior itself is spec compliant, however, it would create
a layout box by default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, a slot does not crate a
layout box by default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
the only use case.

To be precise, non-shadow trees should be read as "non-v1-shadow" trees.

BUG=659761, 660265
Committed: https://crrev.com/6aa048c86241d01d2084127c980d620f39edb8b1
Cr-Commit-Position: refs/heads/master@{#428686}

[hayato, 2016-10-28 08:36:28]

The CQ bit was checked by hayato@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-10-28 08:36:49]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[hayato, 2016-10-28 09:25:10]

Description was changed from

==========
Make slots in non-shadw trees participate in a flat tree

BUG=
==========

to

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Due to maintaining two different life cycle of distributions, we are receiving
crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this behavior is spec compliant in a sense, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

[hayato, 2016-10-28 09:56:34]

hayato@chromium.org changed reviewers:
+ kochi@chromium.org

[hayato, 2016-10-28 09:56:35]

PTAL

[hayato, 2016-10-28 09:57:30]

Description was changed from

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Due to maintaining two different life cycle of distributions, we are receiving
crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this behavior is spec compliant in a sense, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

to

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycle of distributions, we have
been
receiving crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this behavior is spec compliant in a sense, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

[commit-bot: I haz the power, 2016-10-28 09:59:51]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-10-28 09:59:52]

Dry run: Try jobs failed on following builders:
  linux_chromium_rel_ng on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_...)

[hayato, 2016-10-28 10:06:56]

Description was changed from

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycle of distributions, we have
been
receiving crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this behavior is spec compliant in a sense, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

to

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this behavior is spec compliant in a sense, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

[hayato, 2016-10-28 10:07:43]

Description was changed from

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this behavior is spec compliant in a sense, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

to

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this behavior itself is spec compliant, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

[hayato, 2016-10-28 10:07:58]

Description was changed from

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this behavior itself is spec compliant, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

to

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this new behavior itself is spec compliant, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

[hayato, 2016-10-28 10:09:40]

Description was changed from

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this new behavior itself is spec compliant, however, it would create
a layout box in default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box in default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.


BUG=659761,660265
==========

to

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this new behavior itself is spec compliant, however, it would create
a layout box by default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box by default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.

To be precise, non-shadow trees should be read as "non-v1-shadow" trees.

BUG=659761,660265
==========

[kochi, 2016-10-31 05:45:44]

lgtm % nits

Could you mark
imported/csswg-test/css-scoping-1/css-scoping-shadow-slot-fallback.html
as [ Failure ] in LayoutTests/TestExpectations and link
to crbug/660265 or "display: contents" bug?

Until we get display:contents implementation in Blink,
a selector for non-shadow slot can match, and it will
be styled, am I correct?

Nits for subject/description

In Subject:
> Make slots in non-shadw trees participate in a flat tree
s/shadw/shadow/

In description:
> To make the code simple and stable, it would be better to make a slots in

s/a slots/slots/

> case for a slot in a non-shadow tree in a real world.
s/a real world/real world/

> Once Blink supports "display: contents" in the future, slots does not crate a

s/slots/a slot/

> It is likely that cluster fuzz is only the use case.

s/only the/the only/

[hayato, 2016-10-31 07:07:58]

Description was changed from

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in a real world.

To make the code simple and stable, it would be better to make a slots in
non-shadow trees participate in a flat tree, as normal elements.

Note that this new behavior itself is spec compliant, however, it would create
a layout box by default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, slots does not crate a
layout box by default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
only the use case.

To be precise, non-shadow trees should be read as "non-v1-shadow" trees.

BUG=659761,660265
==========

to

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in real world.

To make the code simple and stable, it would be better to make a slot in a
non-shadow tree participate in a flat tree, as normal elements.

Note that this new behavior itself is spec compliant, however, it would create
a layout box by default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, a slot does not crate a
layout box by default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
the only use case.

To be precise, non-shadow trees should be read as "non-v1-shadow" trees.

BUG=659761,660265
==========

[hayato, 2016-10-31 07:09:06]

update expectation

[hayato, 2016-10-31 07:09:07]

The CQ bit was checked by hayato@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-10-31 07:09:21]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[hayato, 2016-10-31 07:12:33]

On 2016/10/31 at 05:45:44, kochi wrote:
> lgtm % nits
> 
> Could you mark
imported/csswg-test/css-scoping-1/css-scoping-shadow-slot-fallback.html
> as [ Failure ] in LayoutTests/TestExpectations and link
> to crbug/660265 or "display: contents" bug?

Nice catch. Done.


> Until we get display:contents implementation in Blink,
> a selector for non-shadow slot can match, and it will
> be styled, am I correct?

Yes, its will be styled.
Note that if Blink supports "display: contents", children of slots will be still
styled, via inheritance from a slot, though slot itself does not generate a
layout box.

> 
> Nits for subject/description
> 
> In Subject:
> > Make slots in non-shadw trees participate in a flat tree
> s/shadw/shadow/
> 
> In description:
> > To make the code simple and stable, it would be better to make a slots in
> 
> s/a slots/slots/
> 
> > case for a slot in a non-shadow tree in a real world.
> s/a real world/real world/
> 
> > Once Blink supports "display: contents" in the future, slots does not crate
a
> 
> s/slots/a slot/
> 
> > It is likely that cluster fuzz is only the use case.
> 
> s/only the/the only/

All done.

[hayato, 2016-10-31 07:13:12]

The CQ bit was unchecked by hayato@chromium.org

[hayato, 2016-10-31 07:13:14]

The CQ bit was checked by hayato@chromium.org

[hayato, 2016-10-31 07:13:15]

The patchset sent to the CQ was uploaded after l-g-t-m from kochi@chromium.org
Link to the patchset: https://codereview.chromium.org/2460813002/#ps20001
(title: "update expectation")

[commit-bot: I haz the power, 2016-10-31 07:13:20]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-10-31 08:09:25]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-10-31 08:09:26]

Try jobs failed on following builders:
  win_chromium_x64_rel_ng on master.tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_x64_...)

[hayato, 2016-10-31 10:16:33]

The CQ bit was checked by hayato@chromium.org

[commit-bot: I haz the power, 2016-10-31 10:16:48]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-10-31 11:32:47]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2016-10-31 11:34:56]

Description was changed from

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in real world.

To make the code simple and stable, it would be better to make a slot in a
non-shadow tree participate in a flat tree, as normal elements.

Note that this new behavior itself is spec compliant, however, it would create
a layout box by default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, a slot does not crate a
layout box by default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
the only use case.

To be precise, non-shadow trees should be read as "non-v1-shadow" trees.

BUG=659761,660265
==========

to

==========
Make slots in non-shadow trees participate in a flat tree

It has required non-trivial efforts to exclude slots in non-shadow trees from a
flat tree so far.  Although this behavior is consistent with that in shadow
trees, this increased code complexity because distributed nodes of a slot in
non-shadow tree are not calculated in a part of normal distribution
calculation flow.

Since it is hard to maintain two different life cycles of distributions, we have
been receiving not-a-few crash reports from cluster fuzz.

The benefit of keeping the current behavior is small; there is no useful use
case for a slot in a non-shadow tree in real world.

To make the code simple and stable, it would be better to make a slot in a
non-shadow tree participate in a flat tree, as normal elements.

Note that this new behavior itself is spec compliant, however, it would create
a layout box by default because Blink does not support "display: contents" yet.
Once Blink supports "display: contents" in the future, a slot does not crate a
layout box by default, even though it participates in a flat tree.

A slot in shadow trees does not participate in a flat tree. This does not
change.

This CL should have no real impacts on users because there is no use cases for
slots in non-shadow trees in real world.  It is likely that cluster fuzz is
the only use case.

To be precise, non-shadow trees should be read as "non-v1-shadow" trees.

BUG=659761,660265

Committed: https://crrev.com/6aa048c86241d01d2084127c980d620f39edb8b1
Cr-Commit-Position: refs/heads/master@{#428686}
==========

[commit-bot: I haz the power, 2016-10-31 11:34:58]

Patchset 2 (id:??) landed as
https://crrev.com/6aa048c86241d01d2084127c980d620f39edb8b1
Cr-Commit-Position: refs/heads/master@{#428686}