[Android] Always set "mobile user agent" switch to the right vlaue when launching the browser proce…

[Android] Always set "mobile user agent" switch to the right vlaue when launching the browser process.

This fixes the user agent for the browser process when the browser process has been started by an activity other than Main, like sync, bookmark provider, etc. If
chrome is started as a background process, the switch is never added. Since the
user agent is computed early on during start-up and not recomputed (from users
of content::GetUserAgent) this can lead to incorrect UA being reported and
broken results.

We probably shouldn't do it this way long-term but this is a minimal fix for m27.

NOTRY=true
BUG=237471
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=199501

[Yaron, 2013-05-10 03:15:20]

Note: this would affect android_webview but I don't think it matters. It's
passed along to the renderer process, but it's only ever read in chrome/ and
content_shell/

[Yaron, 2013-05-10 03:17:31]

crbug.com/230480 (which I've reopened) has more context on why this is happening
more often now.

[bulach, 2013-05-10 08:17:20]

lgtm, but joth is certainly more familiar.

I'd suggest tweaking the description a little bit as well:

"for the browser process when not cold-started to load
a page"
to something like:
"when the browser process has been started by an activity other than Main, like
sync or bookmark provider, or (?)."

[joth, 2013-05-10 09:14:15]

addDeviceSpecificUserAgentSwit**ch() also has the side effect of setting
the "TABLET_UI" flag, if appropriate. [Besides this being a strange
side-effect for a method of that name, and it not sounding like something
that should live in content at all] is that likely to cause webview any
problem? (again, I think it's only used in chrome)

Sounds like we need a chrome/ layer specific wrapper for AndroidBrowserProcess
like we do already for webview (AwBrowserProcess)


lgtm for an m27 quick fix though.



On 10 May 2013 09:17, <bulach@chromium.org> wrote:

> lgtm, but joth is certainly more familiar.
>
> I'd suggest tweaking the description a little bit as well:
>
>
> "for the browser process when not cold-started to load
> a page"
> to something like:
> "when the browser process has been started by an activity other than Main,
> like
> sync or bookmark provider, or (?)."
>
>
https://codereview.chromium.**org/14682021/<https://codereview.chromium.org/1...
>

[Yaron, 2013-05-10 16:39:28]

On 2013/05/10 09:14:15, joth wrote:
> addDeviceSpecificUserAgentSwit**ch() also has the side effect of setting
> the "TABLET_UI" flag, if appropriate. [Besides this being a strange
> side-effect for a method of that name, and it not sounding like something
> that should live in content at all] is that likely to cause webview any
> problem? (again, I think it's only used in chrome)
> 
Yep, I agree and only used in chrome.

> Sounds like we need a chrome/ layer specific wrapper for AndroidBrowserProcess
> like we do already for webview (AwBrowserProcess)
> 
> 
> lgtm for an m27 quick fix though.
> 
Thanks. I'll try and clean this up more afterwards
> 
> 
> On 10 May 2013 09:17, <mailto:bulach@chromium.org> wrote:
> 
> > lgtm, but joth is certainly more familiar.
> >
> > I'd suggest tweaking the description a little bit as well:
> >
> >
> > "for the browser process when not cold-started to load
> > a page"
> > to something like:
> > "when the browser process has been started by an activity other than Main,
> > like
> > sync or bookmark provider, or (?)."
> >
> >
>
https://codereview.chromium.**org/14682021/%3Chttps://codereview.chromium.org...>
> >

[commit-bot: I haz the power, 2013-05-10 17:36:23]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/yfriedman@chromium.org/14682021/1

[commit-bot: I haz the power, 2013-05-10 17:36:51]

Change committed as 199501