Oilpan: fix build after r183657.

Oilpan: fix build after r183657.

TBR=oilpan-reviews,haraken
BUG=
NOTRY=true

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=183666

[sof, 2014-10-14 11:32:25]

Please take a look.

[haraken, 2014-10-14 11:41:30]

https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
File Source/core/rendering/RenderGrid.cpp (right):

https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
Source/core/rendering/RenderGrid.cpp:709: };

Can't we use WTF_ALLOW_INIT_WITH_MEM_FUNCTIONS instead?

[sof, 2014-10-14 11:47:21]

https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
File Source/core/rendering/RenderGrid.cpp (right):

https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
Source/core/rendering/RenderGrid.cpp:709: };
On 2014/10/14 11:41:30, haraken wrote:
> 
> Can't we use WTF_ALLOW_INIT_WITH_MEM_FUNCTIONS instead?

Same as for GradientStop, the reason why not is that we have to override/fix
needsDestruction for this type.

I think we should consider introducing a macro for that, though. Can do that in
a followup, if so.

[haraken, 2014-10-14 11:53:08]

On 2014/10/14 11:47:21, sof wrote:
>
https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
> File Source/core/rendering/RenderGrid.cpp (right):
> 
>
https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
> Source/core/rendering/RenderGrid.cpp:709: };
> On 2014/10/14 11:41:30, haraken wrote:
> > 
> > Can't we use WTF_ALLOW_INIT_WITH_MEM_FUNCTIONS instead?
> 
> Same as for GradientStop, the reason why not is that we have to override/fix
> needsDestruction for this type.
> 
> I think we should consider introducing a macro for that, though. Can do that
in
> a followup, if so.

Just help me understand, what's wrong with calling a destructor of
GridSizingData?


LGTM either way, given I believe you're understanding things correctly.

[sof, 2014-10-14 12:14:27]

On 2014/10/14 11:53:08, haraken wrote:
> On 2014/10/14 11:47:21, sof wrote:
> >
>
https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
> > File Source/core/rendering/RenderGrid.cpp (right):
> > 
> >
>
https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
> > Source/core/rendering/RenderGrid.cpp:709: };
> > On 2014/10/14 11:41:30, haraken wrote:
> > > 
> > > Can't we use WTF_ALLOW_INIT_WITH_MEM_FUNCTIONS instead?
> > 
> > Same as for GradientStop, the reason why not is that we have to override/fix
> > needsDestruction for this type.
> > 
> > I think we should consider introducing a macro for that, though. Can do that
> in
> > a followup, if so.
> 
> Just help me understand, what's wrong with calling a destructor of
> GridSizingData?
> 

(Heap)Vector() doesn't allow finalization of initialized-as-zero elements with
destructors (== non-POD objects), unless canInitializeWithMemset holds. See the
Vector() ctor comment and its assert predicate.

The type here being incorrectly classified as non-POD.

[sof, 2014-10-14 12:14:48]

The CQ bit was checked by sigbjornf@opera.com

[commit-bot: I haz the power, 2014-10-14 12:15:09]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/653053002/1

[commit-bot: I haz the power, 2014-10-14 12:15:41]

Committed patchset #1 (id:1) as 183666

[sof, 2014-10-15 05:16:18]

On 2014/10/14 12:14:27, sof wrote:
> On 2014/10/14 11:53:08, haraken wrote:
> > On 2014/10/14 11:47:21, sof wrote:
> > >
> >
>
https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
> > > File Source/core/rendering/RenderGrid.cpp (right):
> > > 
> > >
> >
>
https://codereview.chromium.org/653053002/diff/1/Source/core/rendering/Render...
> > > Source/core/rendering/RenderGrid.cpp:709: };
> > > On 2014/10/14 11:41:30, haraken wrote:
> > > > 
> > > > Can't we use WTF_ALLOW_INIT_WITH_MEM_FUNCTIONS instead?
> > > 
> > > Same as for GradientStop, the reason why not is that we have to
override/fix
> > > needsDestruction for this type.
> > > 
> > > I think we should consider introducing a macro for that, though. Can do
that
> > in
> > > a followup, if so.
> > 
> > Just help me understand, what's wrong with calling a destructor of
> > GridSizingData?
> > 
> 
> (Heap)Vector() doesn't allow finalization of initialized-as-zero elements with
> destructors (== non-POD objects), unless canInitializeWithMemset holds. See
the
> Vector() ctor comment and its assert predicate.
> 
> The type here being incorrectly classified as non-POD.

Once/when/if we can assume <type_traits> being usable across toolchains (incl
stable wrt final C++11 naming), then this VectorTraits usage can be tidied up a
bit -- https://codereview.chromium.org/654933004/