[web-ui] Migrate RegisterMessageCallback usage to base::bind().

[web-ui] Migrate RegisterMessageCallback usage to base::bind().

(This is a redo of http://codereview.chromium.org/8122004/ which
was reverted.  CrOS breakage has been fixed and will be confirmed by
green trybots.)

BUG=98478
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=103840

[csilv, 2011-10-04 01:35:46]

+jhawkins for review

[James Hawkins, 2011-10-04 01:37:29]

LGTM

[achuithb, 2011-11-22 23:30:00]

I'm not sure base::Unretained is correct here (even though it's effectively what
the previous code did). I think weak ptr should probably be used?

[James Hawkins, 2011-11-22 23:34:42]

On 2011/11/22 23:30:00, achuith.bhandarkar wrote:
> I'm not sure base::Unretained is correct here (even though it's effectively
what
> the previous code did). I think weak ptr should probably be used?

Is this in reference to a particular file:line?

[achuithb, 2011-11-22 23:38:29]

On 2011/11/22 23:34:42, James Hawkins wrote:
> On 2011/11/22 23:30:00, achuith.bhandarkar wrote:
> > I'm not sure base::Unretained is correct here (even though it's effectively
> what
> > the previous code did). I think weak ptr should probably be used?
> 
> Is this in reference to a particular file:line?

I think all the files received the same treatment - I looked at
network_screen_handler.cc, imageburner_ui.cc, user_image_screen_handler.cc.

Is there a guarantee that the callback registered with RegisterMessageCallback
will not come after the handler has been destroyed?

[csilv, 2011-11-23 00:41:16]

On 2011/11/22 23:38:29, achuith.bhandarkar wrote:
> On 2011/11/22 23:34:42, James Hawkins wrote:
> > On 2011/11/22 23:30:00, achuith.bhandarkar wrote:
> > > I'm not sure base::Unretained is correct here (even though it's
effectively
> > what
> > > the previous code did). I think weak ptr should probably be used?
> > 
> > Is this in reference to a particular file:line?
> 
> I think all the files received the same treatment - I looked at
> network_screen_handler.cc, imageburner_ui.cc, user_image_screen_handler.cc.
> 
> Is there a guarantee that the callback registered with RegisterMessageCallback
> will not come after the handler has been destroyed?

Yes, this is guaranteed.  The callbacks are invoked only by the WebUI object
(see WebUI::OnWebUISend).  The WebUI object holds the handlers for the page, and
will delete the handlers only when the WebUI object itself is deleted.

[achuithb, 2011-11-23 00:43:25]

On 2011/11/23 00:41:16, csilv wrote:
> On 2011/11/22 23:38:29, achuith.bhandarkar wrote:
> > On 2011/11/22 23:34:42, James Hawkins wrote:
> > > On 2011/11/22 23:30:00, achuith.bhandarkar wrote:
> > > > I'm not sure base::Unretained is correct here (even though it's
> effectively
> > > what
> > > > the previous code did). I think weak ptr should probably be used?
> > > 
> > > Is this in reference to a particular file:line?
> > 
> > I think all the files received the same treatment - I looked at
> > network_screen_handler.cc, imageburner_ui.cc, user_image_screen_handler.cc.
> > 
> > Is there a guarantee that the callback registered with
RegisterMessageCallback
> > will not come after the handler has been destroyed?
> 
> Yes, this is guaranteed.  The callbacks are invoked only by the WebUI object
> (see WebUI::OnWebUISend).  The WebUI object holds the handlers for the page,
and
> will delete the handlers only when the WebUI object itself is deleted.

Great! Thanks!