Mandoline: Implement basic "process per site" support

Mandoline: Implement basic "process per site" support

After fetching the resource, if it turns out it has a mime type that can
be handled by a known mojo app, then we grab the resources' URL, convert it
into a Site URL and use it as a qualifier to determine what process the mojo
app that will handle the resource should live in.

BUG=496937
Committed: https://crrev.com/d75d12d0827be6d395ac851d37af4ba1ef78b60d
Cr-Commit-Position: refs/heads/master@{#338317}

[Fady Samuel, 2015-07-09 21:20:04]

fsamuel@chromium.org changed reviewers:
+ jam@chromium.org

[Fady Samuel, 2015-07-09 21:20:04]

This probably could use a test or two, but could you please take a look at this
to see if this seems reasonable to you? Thanks!

[jam, 2015-07-09 22:32:37]

thinking about it more, how about having the NetworkFetcher return this to
HandleFetchCallback? that way you don't have to make another call.

https://codereview.chromium.org/1229903003/diff/1/mojo/shell/application_mana...
File mojo/shell/application_manager.cc (right):

https://codereview.chromium.org/1229903003/diff/1/mojo/shell/application_mana...
mojo/shell/application_manager.cc:301: switches::kEnableMultiprocess)) {
why should this behavior change with multiprocess? seems like using qualifier
and having multiple threads of html_viewer when running single process can also
work? the more similar multi-process and single-process, the easier for
testing/maintenance

[Fady Samuel, 2015-07-09 23:10:33]

On 2015/07/09 22:32:37, jam wrote:
> thinking about it more, how about having the NetworkFetcher return this to
> HandleFetchCallback? that way you don't have to make another call.
> 
>
https://codereview.chromium.org/1229903003/diff/1/mojo/shell/application_mana...
> File mojo/shell/application_manager.cc (right):
> 
>
https://codereview.chromium.org/1229903003/diff/1/mojo/shell/application_mana...
> mojo/shell/application_manager.cc:301: switches::kEnableMultiprocess)) {
> why should this behavior change with multiprocess? seems like using qualifier
> and having multiple threads of html_viewer when running single process can
also
> work? the more similar multi-process and single-process, the easier for
> testing/maintenance

Blink doesn't support multiple instances across threads. Its got a lot of
globals.

[Fady Samuel, 2015-07-10 16:04:43]

Patchset #2 (id:20001) has been deleted

[Fady Samuel, 2015-07-10 16:05:45]

PTAL!

[jam, 2015-07-10 16:12:01]

lgtm

https://codereview.chromium.org/1229903003/diff/40001/mojo/services/network/p...
File mojo/services/network/public/interfaces/url_loader.mojom (right):

https://codereview.chromium.org/1229903003/diff/40001/mojo/services/network/p...
mojo/services/network/public/interfaces/url_loader.mojom:51: // The site of the
URL used for determine what process this resource should
nit: since this a generic fetch api, better not to say how this variable is used
by the shell. i.e. just "// The site of the URL."

[Fady Samuel, 2015-07-10 16:16:08]

CQ'ing now.

https://codereview.chromium.org/1229903003/diff/40001/mojo/services/network/p...
File mojo/services/network/public/interfaces/url_loader.mojom (right):

https://codereview.chromium.org/1229903003/diff/40001/mojo/services/network/p...
mojo/services/network/public/interfaces/url_loader.mojom:51: // The site of the
URL used for determine what process this resource should
On 2015/07/10 16:12:00, jam wrote:
> nit: since this a generic fetch api, better not to say how this variable is
used
> by the shell. i.e. just "// The site of the URL."

Done.

[Fady Samuel, 2015-07-10 16:16:11]

The CQ bit was checked by fsamuel@chromium.org

[Fady Samuel, 2015-07-10 16:16:11]

The patchset sent to the CQ was uploaded after l-g-t-m from jam@chromium.org
Link to the patchset: https://codereview.chromium.org/1229903003/#ps60001
(title: "Addressed nit")

[commit-bot: I haz the power, 2015-07-10 16:16:24]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1229903003/60001

[commit-bot: I haz the power, 2015-07-10 17:42:29]

Committed patchset #3 (id:60001)

[commit-bot: I haz the power, 2015-07-10 17:44:07]

Patchset 3 (id:??) landed as
https://crrev.com/d75d12d0827be6d395ac851d37af4ba1ef78b60d
Cr-Commit-Position: refs/heads/master@{#338317}

[jam, 2015-09-10 19:38:47]

hey Fady, ok Darin and I were talking about this. He pointed out that this is a
layering violation, since mojo shell's needs are leaking into URLResponse's
interface.

now that we have perf bots, can we try approach 1 and see if this affects
performance?

Thanks and sorry for the back and forth!