Adds temporary function lastOrientationNew

Adds temporary function lastOrientationNew

This adds a temporary function lastOrientationNew to
device_orientation_dispatcher. This function is identical to lastOrientation.
This will allow us to modify WebKit to use the new function temporarily while
lastOrientation is modified to return a WebDeviceOrientationData instead of a
WebDeviceOrientation. This will allow for consistent naming across WebKit for
DeviceOrientation and DeviceMotion. This change is a follow-up to 
https://bugs.webkit.org/show_bug.cgi?id=88663, which made part of the change
in WebKit.


BUG=none
TEST=none, no change in functionality

[jamesr, 2012-07-25 22:43:12]

I just got back from vacation and it looks like this review is still open - do
you still need a review?  If so, can you verify if this patch still applies to
(roughly) ToT? Thanks!

[aousterh1, 2012-07-31 11:03:56]

Thanks for getting back to me! I do still need a review and I believe that
this patch applies to ToT (when I rebased it on master on my machine there
were no conflicts and all relevant tests still passed). The same is true
for the next patch in the sequence (
https://bugs.webkit.org/show_bug.cgi?id=90972).

Thanks,
-Amy


On Wed, Jul 25, 2012 at 11:43 PM, <jamesr@chromium.org> wrote:

>
> I just got back from vacation and it looks like this review is still open
> - do
> you still need a review?  If so, can you verify if this patch still
> applies to
> (roughly) ToT? Thanks!
>
>
https://chromiumcodereview.**appspot.com/10735052/<https://chromiumcodereview...
>

[jamesr, 2012-08-01 00:40:52]

I don't really get it - what's the reason for adding this function?  It appears
to have the same type and the same name as an existing function, except for the
addition of "new" which doesn't seem to indicate much.

[jamesr, 2012-08-01 00:40:52]

I don't really get it - what's the reason for adding this function?  It appears
to have the same type and the same name as an existing function, except for the
addition of "new" which doesn't seem to indicate much.

[aousterh, 2012-08-01 12:50:59]

On 2012/08/01 00:40:52, jamesr wrote:
> I don't really get it - what's the reason for adding this function?  It
appears
> to have the same type and the same name as an existing function, except for
the
> addition of "new" which doesn't seem to indicate much.

The goal is to rename WebDeviceOrientation to WebDeviceOrientationData in WebKit
(this will be more consistent with other code in WebKit). However,
WebDeviceOrientation is returned by the function lastOrientation in the public
Chromium API in WebKit. If I tried to make this change as one patch in Chromium
and one in WebKit, Chromium would break temporarily until the second patch
landed. Therefore, this needs to be done in several steps:
(1) add lastOrientationNew to Chromium (this function will return a
WebDeviceOrientation) [this Chromium patch]
(2) switch WebKit to call lastOrientationNew, temporarily remove
lastOrientation, and add WebDeviceOrientationData [WebKit patch]
(3) change lastOrientation in Chromium so that it returns a
WebDeviceOrientationData [Chromium patch]
(4) switch WebKit back to calling lastOrientation, remove lastOrientationNew
[WebKit patch]
(5) remove lastOrientationNew and all references to WebDeviceOrientation from
Chromium [Chromium patch]
(6) remove WebDeviceOrientation from WebKit [WebKit patch]

Does that make sense? Do you know of a simpler way of doing it?

[jamesr, 2012-08-01 22:22:32]

On 2012/08/01 12:50:59, aousterh wrote:
> On 2012/08/01 00:40:52, jamesr wrote:
> > I don't really get it - what's the reason for adding this function?  It
> appears
> > to have the same type and the same name as an existing function, except for
> the
> > addition of "new" which doesn't seem to indicate much.
> 
> The goal is to rename WebDeviceOrientation to WebDeviceOrientationData in
WebKit
> (this will be more consistent with other code in WebKit). However,
> WebDeviceOrientation is returned by the function lastOrientation in the public
> Chromium API in WebKit. If I tried to make this change as one patch in
Chromium
> and one in WebKit, Chromium would break temporarily until the second patch
> landed. Therefore, this needs to be done in several steps:
> (1) add lastOrientationNew to Chromium (this function will return a
> WebDeviceOrientation) [this Chromium patch]
> (2) switch WebKit to call lastOrientationNew, temporarily remove
> lastOrientation, and add WebDeviceOrientationData [WebKit patch]
> (3) change lastOrientation in Chromium so that it returns a
> WebDeviceOrientationData [Chromium patch]
> (4) switch WebKit back to calling lastOrientation, remove lastOrientationNew
> [WebKit patch]
> (5) remove lastOrientationNew and all references to WebDeviceOrientation from
> Chromium [Chromium patch]
> (6) remove WebDeviceOrientation from WebKit [WebKit patch]
> 
> Does that make sense? Do you know of a simpler way of doing it?

This seems pretty excessive.  How about this:

1.) Add a version of lastOrientation with the signature you want in chromium
guarded by a #define (something like
LAST_ORIENTATION_RETURNS_WEBDEVICEORIENTATIONDATA) or something like that
alongside the current version [Chromium patch]
2.) Land a WebKit patch that switches to the new return type and sets the
#define in a public WebKit API header [WebKit patch]
3.) After (2) has rolled into Chromium, remove the #ifdef guard and old function
[Chromium patch]
4.) Delete #define from public WebKit API header [WebKit patch]

Far fewer patches and less ugly names.