Add explicit destructors to DOM classes containing smart pointers to incomplete types

Add explicit destructors to DOM classes containing smart pointers to incomplete types

Clang on Windows tries to declare the implicit virtual destructor of
these classes so that it can diagnose mismatched exceptions
specifications. This ends up requiring that the type of the smart
pointer be complete, which it is not. This codepath does not fire on
non-Windows platforms due to the Itanium C++ ABI key function
optimization.

It is possible to fix Clang to avoid performing semantic analysis of the
destructor when it only needs to compute the exception specification,
but it is prohibitively difficult.  http://llvm.org/PR20337

Another possible fix on the Blink side would be to make the smart
pointer type complete by including the relevant header, which has the
cost of increasing transitive header bloat. Instead, simply define the
destructor out of line. This is also a minor object file size
optimization, as now the destructor is only emitted once.

R=thakis@chromium.org
BUG=82385
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=178838

[Reid Kleckner, 2014-07-23 02:32:23]

This is a re-send of with a fixed blink baseurl, so that I can get green try
jobs.  PTAL

[Nico, 2014-07-23 04:07:40]

rslgtm, but are you sure about the base url? i think both should probably work.
I ran `git cl patch 412483003`and got this output:

$ git cl patch 407313003
Loaded authentication cookies from /Users/thakis/.codereview_upload_cookies
error: patch failed: Source/core/html/HTMLOptionElement.cpp:54
Falling back to three-way merge...
Applied patch to 'Source/core/html/HTMLOptionElement.cpp' cleanly.
error: patch failed: Source/core/html/HTMLOptionElement.h:70
Falling back to three-way merge...
Applied patch to 'Source/core/html/HTMLOptionElement.h' cleanly.
Committed patch locally.

…so maybe you're just too far behind trunk? (Maybe you're using the blink
revision that's deps'd into chromium instead of blink head?)

You can do `git log` in your blink checkout and then `git cl try --revision R`
with the revision you have to make sure your patch applies cleanly.

This patch and your other patch have the same errors.

( https://codereview.chromium.org/409333003 is this patch applied to trunk blink
with try jobs from there. So if that one goes green, you can rebase either of
your two patches to trunk and then just land it.)

[Reid Kleckner, 2014-07-24 00:04:01]

The CQ bit was checked by rnk@chromium.org

[commit-bot: I haz the power, 2014-07-24 00:04:46]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/rnk@chromium.org/412483007/20001

[commit-bot: I haz the power, 2014-07-24 06:41:51]

FYI, CQ is re-trying this CL (attempt #1).
The failing builders are:
  win_gpu_triggered_tests on tryserver.chromium.gpu
(http://build.chromium.org/p/tryserver.chromium.gpu/builders/win_gpu_triggered...)

[commit-bot: I haz the power, 2014-07-24 07:07:22]

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

[commit-bot: I haz the power, 2014-07-24 07:07:23]

Try jobs failed on following builders:
  win_gpu_triggered_tests on tryserver.chromium.gpu
(http://build.chromium.org/p/tryserver.chromium.gpu/builders/win_gpu_triggered...)

[Nico, 2014-07-24 13:34:17]

The CQ bit was checked by thakis@chromium.org

[commit-bot: I haz the power, 2014-07-24 13:35:23]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/rnk@chromium.org/412483007/20001

[commit-bot: I haz the power, 2014-07-24 13:37:03]

Change committed as 178838

[Nico, 2015-01-27 19:42:49]

A revert of this CL (patchset #2 id:20001) has been created in
https://codereview.chromium.org/878273002/ by thakis@chromium.org.

The reason for reverting is: This should no longer be necessary; clang should be
able to handle this now. Reverting this to test that theory..