Cleanup: make ExtensionRequestSender manage its IPC responses directly, rather

Cleanup: make ExtensionRequestSender manage its IPC responses directly, rather
than going via ExtensionHelper and ExtensionDispatcher. Also make it generate
its request IDs internally.

A cleanup that leads from this is making ExtensionMsg_Reponse a CONTROL message
rather than ROUTED, simplifying some browser-side code.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=145642

[koz (OOO until 15th September), 2012-07-09 01:12:37]

http://codereview.chromium.org/10703111/diff/2003/chrome/renderer/resources/e...
File chrome/renderer/resources/extensions/send_request.js (right):

http://codereview.chromium.org/10703111/diff/2003/chrome/renderer/resources/e...
chrome/renderer/resources/extensions/send_request.js:118: var requestId =
nativeFunction(functionName,
set_icon.js provides a nativeFunction() to be called here, but I don't see a
change for that file in this CL.

[not at google - send to devlin, 2012-07-09 02:04:16]

http://codereview.chromium.org/10703111/diff/2003/chrome/renderer/resources/e...
File chrome/renderer/resources/extensions/send_request.js (right):

http://codereview.chromium.org/10703111/diff/2003/chrome/renderer/resources/e...
chrome/renderer/resources/extensions/send_request.js:118: var requestId =
nativeFunction(functionName,
On 2012/07/09 01:12:37, koz wrote:
> set_icon.js provides a nativeFunction() to be called here, but I don't see a
> change for that file in this CL.

Thanks for noticing. I just checked and set_icon.js just calls directly into
set_icon_natives.cc and that I *did* remember to update (well, C++ did cos it
has *types* woo!).

[koz (OOO until 15th September), 2012-07-09 02:18:12]

lgtm

[commit-bot: I haz the power, 2012-07-09 02:37:50]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/10703111/4002

[commit-bot: I haz the power, 2012-07-09 03:53:34]

Change committed as 145642

[cmp_google, 2012-07-09 21:18:50]

Hi all, when this landed, it made some of our tests go red that use the
reference build:
http://build.chromium.org/p/chromium.perf/builders/Win7%20Interactive%20Perf/...

Is there a way to land this change so that tests compiled using the new code
will remain compatible with the reference build?  If not, my assumption is that
this change will require a reference build update if you want to land it. 
Please grab me on IM to discuss more.  For now, I plan to revert this change to
see if these builds go back to green and to make sure we know this CL caused the
failure.

[kalman_google, 2012-07-10 01:15:28]

Ah, bummer. I didn't realise we did this.

I guess the problem is that I change the IPC format? Though I thought that
my change was more or less a no-op in this regard, since CONTROL is just
ROUTED with one of the necessary parameters curried in already.

Anyway, what's the architecture of these tests? We... compile tests then
run against the old binary, how does that work?

I would ideally like to land this change, though it's just a cleanup.


On Tue, Jul 10, 2012 at 7:18 AM, <[email protected]> wrote:

> Hi all, when this landed, it made some of our tests go red that use the
> reference build:
> http://build.chromium.org/p/**chromium.perf/builders/Win7%**
>
20Interactive%20Perf/builds/**585/<http://build.chromium.org/p/chromium.perf/builders/Win7%20Interactive%20Perf/builds/585/>
>
> Is there a way to land this change so that tests compiled using the new
> code
> will remain compatible with the reference build?  If not, my assumption is
> that
> this change will require a reference build update if you want to land it.
> Please grab me on IM to discuss more.  For now, I plan to revert this
> change to
> see if these builds go back to green and to make sure we know this CL
> caused the
> failure.
>
>
https://chromiumcodereview.**appspot.com/10703111/<https://chromiumcodereview...
>

[cmp, 2012-07-12 21:07:31]

On 2012/07/10 01:15:28, kalman_google wrote:
> Anyway, what's the architecture of these tests? We... compile tests then
> run against the old binary, how does that work?

Right, the current Chrome build exposes the same IPC automation interface as the
reference Chrome build.  This works most of the time. :)  It happens more often
that we need a reference build update just because the IPC automation messages
change by itself.  It is less common that a new feature is added to Chrome that
changes something about what we test that requires a reference build update.

> I would ideally like to land this change, though it's just a cleanup.

If you want to do that and you're also willing to update the reference build
plus assist in rebaselining the perf_expectations.json file, let me know.