Mojo C++ Bindings: Introduce mojo::SupportsStrongBinding

Mojo C++ Bindings: Introduce mojo::SupportsStrongBinding

Adds mojo::SupportsStrongBinding and requires this to be
inherited by any interface implementation that will be
strongly bound to a pipe.

This allows for safe deletion of strongly bound impls
from outside their pipe's connection error handler.

BUG=666803
R=dcheng@chromium.org

[Ken Rockot(use gerrit already), 2016-11-19 06:48:59]

Description was changed from

==========
Mojo C++ Bindings: Introduce mojo::SupportsStrongBinding

Adds mojo::SupportsStrongBinding and requires this to be
inherited by any interface implementation that will be
strongly bound to a pipe.

This allows for safe deletion of the strongly bound impls
from outside their pipe's connection error handler.

BUG=666803
R=dcheng@chromium.org

patch from issue 2504193006 at patchset 20001
(http://crrev.com/2504193006#ps20001)
==========

to

==========
Mojo C++ Bindings: Introduce mojo::SupportsStrongBinding

Adds mojo::SupportsStrongBinding and requires this to be
inherited by any interface implementation that will be
strongly bound to a pipe.

This allows for safe deletion of strongly bound impls
from outside their pipe's connection error handler.

BUG=666803
R=dcheng@chromium.org

patch from issue 2504193006 at patchset 20001
(http://crrev.com/2504193006#ps20001)
==========

[Ken Rockot(use gerrit already), 2016-11-19 06:54:36]

Description was changed from

==========
Mojo C++ Bindings: Introduce mojo::SupportsStrongBinding

Adds mojo::SupportsStrongBinding and requires this to be
inherited by any interface implementation that will be
strongly bound to a pipe.

This allows for safe deletion of strongly bound impls
from outside their pipe's connection error handler.

BUG=666803
R=dcheng@chromium.org

patch from issue 2504193006 at patchset 20001
(http://crrev.com/2504193006#ps20001)
==========

to

==========
Mojo C++ Bindings: Introduce mojo::SupportsStrongBinding

Adds mojo::SupportsStrongBinding and requires this to be
inherited by any interface implementation that will be
strongly bound to a pipe.

This allows for safe deletion of strongly bound impls
from outside their pipe's connection error handler.

BUG=666803
R=dcheng@chromium.org
==========

[Ken Rockot(use gerrit already), 2016-11-19 06:54:40]

Patchset #1 (id:1) has been deleted

[Ken Rockot(use gerrit already), 2016-11-19 06:55:24]

Here's what it might look like to make self deletion safe.

[dcheng, 2016-11-21 10:42:12]

Two thoughts:
1) If we consider this route, can we just embed the StrongBinding itself in
SupportsStrongBinding? I think that would remove the need for the
x->Release().release() call.

2) I guess this comes down to the age-old debate of inheritance vs composition.
But in this sense, I think it makes sense to use inheritance... hopefully I
don't rue these words in the future =)

https://codereview.chromium.org/2515873003/diff/20001/mojo/public/cpp/binding...
File mojo/public/cpp/bindings/strong_binding.h (right):

https://codereview.chromium.org/2515873003/diff/20001/mojo/public/cpp/binding...
mojo/public/cpp/bindings/strong_binding.h:79: static StrongBindingPtr<Interface>
Create(
Would this still need to return a weak pointer to the strong binding once we fix
up existing usage of this return value?

I think we wouldn't need it with this model?

[Ken Rockot(use gerrit already), 2016-11-21 15:40:13]

On Nov 21, 2016 2:42 AM, <dcheng@chromium.org> wrote:
>
> Two thoughts:
> 1) If we consider this route, can we just embed the StrongBinding itself
in
> SupportsStrongBinding? I think that would remove the need for the
> x->Release().release() call.

I would like to, but see below.

>
> 2) I guess this comes down to the age-old debate of inheritance vs
composition.
> But in this sense, I think it makes sense to use inheritance... hopefully
I
> don't rue these words in the future =)

Yes, it makes me sad to resort to inheritance, but if we want to meet the
constraints of having clearer memory management while making it safe to
write intuitive code, I don't see a better option.

>
>
>
https://codereview.chromium.org/2515873003/diff/20001/mojo/public/cpp/binding...
> File mojo/public/cpp/bindings/strong_binding.h (right):
>
>
https://codereview.chromium.org/2515873003/diff/20001/mojo/public/cpp/binding...
> mojo/public/cpp/bindings/strong_binding.h:79: static
> StrongBindingPtr<Interface> Create(
> Would this still need to return a weak pointer to the strong binding
> once we fix up existing usage of this return value?
>
> I think we wouldn't need it with this model?
>

It's useful for tests, but probably not much else. Maybe we can use Binding
internally, and exposing it via SupportsStrongBinding from a testing-only
interface.

> https://codereview.chromium.org/2515873003/

-- 
You received this message because you are subscribed to the Google Groups
"Chromium-reviews" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[dcheng, 2016-12-09 05:27:58]

Note to myself: useful to figure out if we can handle cases like
https://codereview.chromium.org/2474653003/diff/340001/chrome/browser/prefs/p...
more elegantly as well.