First patch in making destructors of refcounted objects private.

First patch in making destructors of refcounted objects private.

BUG=26749
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=31136

[M-A Ruel, 2009-11-05 20:31:28]

lgtm with a few fixes.

http://codereview.chromium.org/360042/diff/1/17
File base/field_trial.h (right):

http://codereview.chromium.org/360042/diff/1/17#newcode121
Line 121: virtual ~FieldTrial() {}
why virtual?

http://codereview.chromium.org/360042/diff/1/21
File ipc/ipc_channel_proxy.h (right):

http://codereview.chromium.org/360042/diff/1/21#newcode172
Line 172: friend class base::RefCountedThreadSafe<Context>;
protected?

http://codereview.chromium.org/360042/diff/1/20
File ipc/ipc_sync_channel.h (right):

http://codereview.chromium.org/360042/diff/1/20#newcode98
Line 98: ~SyncContext();
virtual

http://codereview.chromium.org/360042/diff/1/20#newcode102
Line 102: virtual void Clear();
please fix alignment while at it.

http://codereview.chromium.org/360042/diff/1/4
File media/base/buffers.h (right):

http://codereview.chromium.org/360042/diff/1/4#newcode122
Line 122: ~WritableBuffer() {}
virtual

http://codereview.chromium.org/360042/diff/1/5
File media/base/factory.h (right):

http://codereview.chromium.org/360042/diff/1/5#newcode85
Line 85: MediaFilter* Create(FilterType filter_type, const MediaFormat&
media_format) {
virtual

http://codereview.chromium.org/360042/diff/1/5#newcode96
Line 96: ~FilterFactoryCollection() {}
virtual

http://codereview.chromium.org/360042/diff/1/5#newcode128
Line 128: ~FilterFactoryImpl0() {}
virtual

[jam, 2009-11-05 21:52:37]

http://codereview.chromium.org/360042/diff/1/17
File base/field_trial.h (right):

http://codereview.chromium.org/360042/diff/1/17#newcode121
Line 121: virtual ~FieldTrial() {}
On 2009/11/05 20:31:28, Marc-Antoine Ruel wrote:
> why virtual?

Done.

http://codereview.chromium.org/360042/diff/1/21
File ipc/ipc_channel_proxy.h (right):

http://codereview.chromium.org/360042/diff/1/21#newcode172
Line 172: friend class base::RefCountedThreadSafe<Context>;
On 2009/11/05 20:31:28, Marc-Antoine Ruel wrote:
> protected?

yeah, because SyncContext derives from it.

http://codereview.chromium.org/360042/diff/1/20
File ipc/ipc_sync_channel.h (right):

http://codereview.chromium.org/360042/diff/1/20#newcode98
Line 98: ~SyncContext();
On 2009/11/05 20:31:28, Marc-Antoine Ruel wrote:
> virtual

no need, nothing derives from this class.

http://codereview.chromium.org/360042/diff/1/20#newcode102
Line 102: virtual void Clear();
On 2009/11/05 20:31:28, Marc-Antoine Ruel wrote:
> please fix alignment while at it.

Done.

http://codereview.chromium.org/360042/diff/1/4
File media/base/buffers.h (right):

http://codereview.chromium.org/360042/diff/1/4#newcode122
Line 122: ~WritableBuffer() {}
On 2009/11/05 20:31:28, Marc-Antoine Ruel wrote:
> virtual

Done.

http://codereview.chromium.org/360042/diff/1/5
File media/base/factory.h (right):

http://codereview.chromium.org/360042/diff/1/5#newcode96
Line 96: ~FilterFactoryCollection() {}
On 2009/11/05 20:31:28, Marc-Antoine Ruel wrote:
> virtual

no child classes.  if they get added in the future, they'll get a compile time
error and can mark this as non-virtual.  before there was no destructor
specified anyways, so this matches the same behavior

[M-A Ruel, 2009-11-05 21:56:04]

ok, please commit. :)

[M-A Ruel, 2009-11-05 21:56:23]

On 2009/11/05 21:56:04, Marc-Antoine Ruel wrote:
> ok, please commit. :)

oops, was already committed. I wonder when you found the time, the tree was
always closed today...

[awong, 2009-11-05 21:58:39]

drive-by comment.

http://codereview.chromium.org/360042/diff/1/20
File ipc/ipc_sync_channel.h (right):

http://codereview.chromium.org/360042/diff/1/20#newcode98
Line 98: ~SyncContext();
On 2009/11/05 21:52:37, John Abd-El-Malek wrote:
> On 2009/11/05 20:31:28, Marc-Antoine Ruel wrote:
> > virtual
> 
> no need, nothing derives from this class.

It's already virtual because it's inheriting from a class that has a virtual
destructor.  Making it virtual here is just good documentation.  (Also the style
guide says to. :) )

Same with the other childless classes that are descendants of others with
virtual destructors.