Chromium roll (https://codereview.chromium.org/976353002)

Update from https://crrev.com/319330
- New chromium clang rules require explicit external destructors so
  system/lib added for MessagePipe, DataPipe and SharedBuffer
- New chromium clang rules require override and no virtual in
  declarations, so many files updated.
- cc_strip_video patch updated.

BUG=
R=jamesr@chromium.org

Committed: https://chromium.googlesource.com/external/mojo/+/cc0e4f9d4e73866f10c47b46f2a369df4fd24ce9

[DaveMoore, 2015-03-06 21:19:22]

Fix Clipboard.java

[jamesr, 2015-03-06 22:35:32]

In your patch description clank != clang

for the GN error on nacl, see
https://groups.google.com/a/chromium.org/forum/#!searchin/mojo-dev/breaking/m...

https://codereview.chromium.org/988693005/diff/10001/mojo/public/cpp/system/B...
File mojo/public/cpp/system/BUILD.gn (right):

https://codereview.chromium.org/988693005/diff/10001/mojo/public/cpp/system/B...
mojo/public/cpp/system/BUILD.gn:14: "lib/buffer.cc",
i think we should keep this target header-only. could you disable the clang
checker for the mojo repo while we figure out how to resolve the d'tor issue?

https://codereview.chromium.org/988693005/diff/10001/services/reaper/reaper_b...
File services/reaper/reaper_binding.cc (right):

https://codereview.chromium.org/988693005/diff/10001/services/reaper/reaper_b...
services/reaper/reaper_binding.cc:17: ReaperBinding::~ReaperBinding() {}
i know you had to out-of-line this but why change it from =default to explicit?
was there a particular reason the previous code was defaulted (I think you
reviewed this, right)?

[DaveMoore, 2015-03-06 23:31:48]

Turn off find-bad-constructs and use inline destructors

[DaveMoore, 2015-03-06 23:35:28]

Revert {reaper,transfer}_binding to inline destructors

[DaveMoore, 2015-03-09 18:23:25]

cpu_arch -> target_cpu

[DaveMoore, 2015-03-09 18:36:01]

Rev DEPS.nacl revision

[jamesr, 2015-03-10 18:09:05]

jamesr@chromium.org changed reviewers:
+ jamesr@chromium.org

[jamesr, 2015-03-10 18:09:07]

The reason this fails on the trybots only is the order the trybots process
patches is

(1) Apply only DEPS change in patch
(2) gclient sync
(3) Apply all changes in patch except for DEPS
(4) proceed to next steps

In particular, there's no 'gclient sync' after step (3).  On this patch the
'DEPS.nacl' change is *not* applied before running 'gclient sync' so the trybot
is trying to use the pre-patched version of native_client.

[DaveMoore, 2015-03-10 20:37:00]

Android build fix

[DaveMoore, 2015-03-10 21:24:00]

The tries are all failing because the patch is too big so I'm running all the
configs by hand. PTAL at the cl itself.

https://codereview.chromium.org/988693005/diff/10001/mojo/public/cpp/system/B...
File mojo/public/cpp/system/BUILD.gn (right):

https://codereview.chromium.org/988693005/diff/10001/mojo/public/cpp/system/B...
mojo/public/cpp/system/BUILD.gn:14: "lib/buffer.cc",
On 2015/03/06 22:35:32, jamesr wrote:
> i think we should keep this target header-only. could you disable the clang
> checker for the mojo repo while we figure out how to resolve the d'tor issue?

Done.

https://codereview.chromium.org/988693005/diff/10001/services/reaper/reaper_b...
File services/reaper/reaper_binding.cc (right):

https://codereview.chromium.org/988693005/diff/10001/services/reaper/reaper_b...
services/reaper/reaper_binding.cc:17: ReaperBinding::~ReaperBinding() {}
On 2015/03/06 22:35:32, jamesr wrote:
> i know you had to out-of-line this but why change it from =default to
explicit?
> was there a particular reason the previous code was defaulted (I think you
> reviewed this, right)?

Done.

[jamesr, 2015-03-10 21:42:14]

lgtm

https://codereview.chromium.org/988693005/diff/60001/mojo/tools/roll/disable_...
File mojo/tools/roll/disable_find_bad_constructs.patch (right):

https://codereview.chromium.org/988693005/diff/60001/mojo/tools/roll/disable_...
mojo/tools/roll/disable_find_bad_constructs.patch:8: "-Xclang",
what this block used to do was add

"-Xclang -add-plugin -Xclang find-bad-constructs"

The -Xclang prefixes said to pass along the command to clang, so this really
meant

"-add-plugin find-bad-constructs"

you've now changed this to say

"-Xclang -add-plugin -Xclang"

so whatever argument comes after this will be misinterpreted as a plugin to
attempt to add.  That's not what you want I'm sure. instead you should patch
this to not add anything.

[DaveMoore, 2015-03-10 22:22:46]

fixed bad android build patch

[DaveMoore, 2015-03-10 22:23:11]

Committed patchset #8 (id:70001) manually as
cc0e4f9d4e73866f10c47b46f2a369df4fd24ce9.