Revert of WTF::bind: Handle movable objects in unbound arguments.

Revert of WTF::bind: Handle movable objects in unbound arguments. (patchset #1 id:1 of https://codereview.chromium.org/1744143002/ )

Reason for revert:
Speculative revert to see if it fixes leaks in WebKit tests on Linux that started when this CL was checked in:

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Leak/builds/17815

Only two other changes (besides this CL) went into that build (one being Mac only) and this CL happens to be the only checkin to WebKit directly. Will reland if not the culprit.

Original issue's description:
> WTF::bind: Handle movable objects in unbound arguments.
>
> This patch is the complement to an earlier patch
> (https://codereview.chromium.org/1708183002/), which did the same thing
> for bound arguments.
>
> Fix for unbound arguments is simpler than that for bound arguments. We
> just need to make sure the arguments are passed correctly throughout the
> process.
>
> Due to the current implementation where we use a virtual function to
> implement PartBoundFunctionImpl::operator() we need to copy- or
> move-construct a value there, if the argument's type is not a
> reference. This is extra cost, but it's probably not feasible to get rid
> of this overhead with the current design.
>
> More comments and tests are added accordingly.
>
> BUG=565765
> R=haraken@chromium.org, hiroshige@chromium.org, tzik@chromium.org
>
> Committed: https://crrev.com/cbf29aa70a0a46e631f77e6b6256a3771a09b994
> Cr-Commit-Position: refs/heads/master@{#378372}

TBR=haraken@chromium.org,hiroshige@chromium.org,tzik@chromium.org,mikhail.pozdnyakov@intel.com,yutak@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=565765
Committed: https://crrev.com/a55ac737b7029b9057bc021dce6c1ef3788b212e
Cr-Commit-Position: refs/heads/master@{#378746}

[Finnur, 2016-03-02 11:47:29]

Created Revert of WTF::bind: Handle movable objects in unbound arguments.

[commit-bot: I haz the power, 2016-03-02 11:47:38]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1757873002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1757873002/1

[commit-bot: I haz the power, 2016-03-02 12:01:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1757873002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1757873002/1

[commit-bot: I haz the power, 2016-03-02 12:31:17]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1757873002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1757873002/1

[commit-bot: I haz the power, 2016-03-02 13:01:09]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1757873002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1757873002/1

[commit-bot: I haz the power, 2016-03-02 13:31:24]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1757873002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1757873002/1

[commit-bot: I haz the power, 2016-03-02 14:01:19]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1757873002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1757873002/1

[commit-bot: I haz the power, 2016-03-02 14:31:29]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1757873002/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1757873002/1

[commit-bot: I haz the power, 2016-03-02 14:36:57]

Description was changed from

==========
Revert of WTF::bind: Handle movable objects in unbound arguments. (patchset #1
id:1 of https://codereview.chromium.org/1744143002/ )

Reason for revert:
Speculative revert to see if it fixes leaks in WebKit tests on Linux that
started when this CL was checked in:

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Leak/b...

Only two other changes (besides this CL) went into that build (one being Mac
only) and this CL happens to be the only checkin to WebKit directly. Will reland
if not the culprit.

Original issue's description:
> WTF::bind: Handle movable objects in unbound arguments.
> 
> This patch is the complement to an earlier patch
> (https://codereview.chromium.org/1708183002/), which did the same thing
> for bound arguments.
> 
> Fix for unbound arguments is simpler than that for bound arguments. We
> just need to make sure the arguments are passed correctly throughout the
> process.
> 
> Due to the current implementation where we use a virtual function to
> implement PartBoundFunctionImpl::operator() we need to copy- or
> move-construct a value there, if the argument's type is not a
> reference. This is extra cost, but it's probably not feasible to get rid
> of this overhead with the current design.
> 
> More comments and tests are added accordingly.
> 
> BUG=565765
> R=haraken@chromium.org, hiroshige@chromium.org, tzik@chromium.org
> 
> Committed: https://crrev.com/cbf29aa70a0a46e631f77e6b6256a3771a09b994
> Cr-Commit-Position: refs/heads/master@{#378372}

TBR=haraken@chromium.org,hiroshige@chromium.org,tzik@chromium.org,mikhail.poz...
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=565765
==========

to

==========
Revert of WTF::bind: Handle movable objects in unbound arguments. (patchset #1
id:1 of https://codereview.chromium.org/1744143002/ )

Reason for revert:
Speculative revert to see if it fixes leaks in WebKit tests on Linux that
started when this CL was checked in:

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Leak/b...

Only two other changes (besides this CL) went into that build (one being Mac
only) and this CL happens to be the only checkin to WebKit directly. Will reland
if not the culprit.

Original issue's description:
> WTF::bind: Handle movable objects in unbound arguments.
> 
> This patch is the complement to an earlier patch
> (https://codereview.chromium.org/1708183002/), which did the same thing
> for bound arguments.
> 
> Fix for unbound arguments is simpler than that for bound arguments. We
> just need to make sure the arguments are passed correctly throughout the
> process.
> 
> Due to the current implementation where we use a virtual function to
> implement PartBoundFunctionImpl::operator() we need to copy- or
> move-construct a value there, if the argument's type is not a
> reference. This is extra cost, but it's probably not feasible to get rid
> of this overhead with the current design.
> 
> More comments and tests are added accordingly.
> 
> BUG=565765
> R=haraken@chromium.org, hiroshige@chromium.org, tzik@chromium.org
> 
> Committed: https://crrev.com/cbf29aa70a0a46e631f77e6b6256a3771a09b994
> Cr-Commit-Position: refs/heads/master@{#378372}

TBR=haraken@chromium.org,hiroshige@chromium.org,tzik@chromium.org,mikhail.poz...
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=565765
==========

[commit-bot: I haz the power, 2016-03-02 14:36:58]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-03-02 14:38:21]

Description was changed from

==========
Revert of WTF::bind: Handle movable objects in unbound arguments. (patchset #1
id:1 of https://codereview.chromium.org/1744143002/ )

Reason for revert:
Speculative revert to see if it fixes leaks in WebKit tests on Linux that
started when this CL was checked in:

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Leak/b...

Only two other changes (besides this CL) went into that build (one being Mac
only) and this CL happens to be the only checkin to WebKit directly. Will reland
if not the culprit.

Original issue's description:
> WTF::bind: Handle movable objects in unbound arguments.
> 
> This patch is the complement to an earlier patch
> (https://codereview.chromium.org/1708183002/), which did the same thing
> for bound arguments.
> 
> Fix for unbound arguments is simpler than that for bound arguments. We
> just need to make sure the arguments are passed correctly throughout the
> process.
> 
> Due to the current implementation where we use a virtual function to
> implement PartBoundFunctionImpl::operator() we need to copy- or
> move-construct a value there, if the argument's type is not a
> reference. This is extra cost, but it's probably not feasible to get rid
> of this overhead with the current design.
> 
> More comments and tests are added accordingly.
> 
> BUG=565765
> R=haraken@chromium.org, hiroshige@chromium.org, tzik@chromium.org
> 
> Committed: https://crrev.com/cbf29aa70a0a46e631f77e6b6256a3771a09b994
> Cr-Commit-Position: refs/heads/master@{#378372}

TBR=haraken@chromium.org,hiroshige@chromium.org,tzik@chromium.org,mikhail.poz...
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=565765
==========

to

==========
Revert of WTF::bind: Handle movable objects in unbound arguments. (patchset #1
id:1 of https://codereview.chromium.org/1744143002/ )

Reason for revert:
Speculative revert to see if it fixes leaks in WebKit tests on Linux that
started when this CL was checked in:

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Leak/b...

Only two other changes (besides this CL) went into that build (one being Mac
only) and this CL happens to be the only checkin to WebKit directly. Will reland
if not the culprit.

Original issue's description:
> WTF::bind: Handle movable objects in unbound arguments.
>
> This patch is the complement to an earlier patch
> (https://codereview.chromium.org/1708183002/), which did the same thing
> for bound arguments.
>
> Fix for unbound arguments is simpler than that for bound arguments. We
> just need to make sure the arguments are passed correctly throughout the
> process.
>
> Due to the current implementation where we use a virtual function to
> implement PartBoundFunctionImpl::operator() we need to copy- or
> move-construct a value there, if the argument's type is not a
> reference. This is extra cost, but it's probably not feasible to get rid
> of this overhead with the current design.
>
> More comments and tests are added accordingly.
>
> BUG=565765
> R=haraken@chromium.org, hiroshige@chromium.org, tzik@chromium.org
>
> Committed: https://crrev.com/cbf29aa70a0a46e631f77e6b6256a3771a09b994
> Cr-Commit-Position: refs/heads/master@{#378372}

TBR=haraken@chromium.org,hiroshige@chromium.org,tzik@chromium.org,mikhail.poz...
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG=565765

Committed: https://crrev.com/a55ac737b7029b9057bc021dce6c1ef3788b212e
Cr-Commit-Position: refs/heads/master@{#378746}
==========

[commit-bot: I haz the power, 2016-03-02 14:38:22]

Patchset 1 (id:??) landed as
https://crrev.com/a55ac737b7029b9057bc021dce6c1ef3788b212e
Cr-Commit-Position: refs/heads/master@{#378746}

[Yuta Kitamura, 2016-03-03 04:35:22]

On 2016/03/02 11:47:29, Finnur wrote:
> Created Revert of WTF::bind: Handle movable objects in unbound arguments.

finnur: My guess is that the revert did not work, but the bot
history is somewhat complicated, so I'd like to defer the final
decision to you. If you think my theory is plausible, feel free
to reland my patch.

Given the following history, I think my patch is irrelevant to
the mega leak (leaks in 300+ tests) or service worker related leak.

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Leak/b...
  --> 393 unexpected leaks

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Leak/b...
  --> 25 unexpected leaks (mostly service worker related)
      == Some patch between job 17837 and 17838 appeared to fix the mega leak

...

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Leak/b...
  --> 24 unexpected leaks

https://build.chromium.org/p/chromium.webkit/builders/WebKit%20Linux%20Leak/b...
  --> The revert was landed here, but still 25 unexpected leaks remaining
      == The revert did not fix the leak

[Finnur, 2016-03-03 10:40:31]

A revert of this CL (patchset #1 id:1) has been created in
https://codereview.chromium.org/1760133002/ by finnur@chromium.org.

The reason for reverting is: Re-landing CL since it didn't fix the leaks..