crbug.com/127841 - Request Tablet Site on CB with touch screen.

chrome/browser/ui/browser_commands.cc

Index: chrome/browser/ui/browser_commands.cc
diff --git a/chrome/browser/ui/browser_commands.cc b/chrome/browser/ui/browser_commands.cc
index 423efc915ea14d1bdfdb860592384f17b6eef198..c267d783356a4630ee7e80bfdda3e877ab5db3ad 100644
--- a/chrome/browser/ui/browser_commands.cc
+++ b/chrome/browser/ui/browser_commands.cc
@@ -62,10 +62,12 @@
 #include "content/public/browser/user_metrics.h"
 #include "content/public/browser/web_contents.h"
 #include "content/public/browser/web_contents_view.h"
+#include "content/public/common/content_client.h"
 #include "content/public/common/content_restriction.h"
 #include "content/public/common/renderer_preferences.h"
 #include "content/public/common/url_constants.h"
 #include "net/base/escape.h"
+#include "webkit/glue/user_agent.h"
 #include "webkit/glue/webkit_glue.h"
 
 #if defined(OS_MACOSX)
@@ -864,6 +866,24 @@ void ToggleSpeechInput(Browser* browser) {
   GetActiveWebContents(browser)->GetRenderViewHost()->ToggleSpeechInput();
 }
 
+void ToggleRequestTabletSite(Browser* browser)
+{
+  WebContents* current_tab = GetActiveWebContents(browser);
+  if (!current_tab) return;
+  NavigationEntry *entry = current_tab->GetController().GetActiveEntry();
+  if (!entry) return;
+  if (entry->GetIsOverridingUserAgent()) {
+    entry->SetIsOverridingUserAgent(false);
+    current_tab->SetUserAgentOverride(std::string());

[gone, 2012/08/03 00:08:26]

I don't think you have to worry about un-setting the user agent override string
here, unless you want to be doubly sure that the string doesn't get used. 
Setting the boolean to false should probably be enough.

Also, if this is going to persist as a tab-level setting, you should iterate
over all of the NavigationEntrys in the NavigationController and set the
override flags accordingly.

[sschmitz, 2012/08/03 15:36:58]

I agree, but I prefer to unset the string. It seems a bit more as defensive
programming.
Done

[Rick Byers, 2012/08/03 15:53:06]

I'm not sure that's what we want.  This is about scenarios like the following,
right?
1. Go to a site
2. Click on a link
3. Turn on 'request tablet site'
4. Click back

In that case I think we should be back in the state where the setting is off -
just like Android.  I may have over-simplified earlier saying that the setting
should persist for the life of the tab.  I meant just that when it's enabled and
you enter a new URL in the omnibox, it would still be enabled.  I think it would
be confusing for the user if the state wasn't preserved with back/forward.

[sschmitz, 2012/08/03 18:11:43]

Yes, good question. I can see benefits for both scenarios "per tab (all
entries)" vs. "per current entry only". I'm leaning towards the "Per Tab (for
all entries)" scenario. Once the user toggles "RTS" to on, it "sticks" for the
current tab. I think this may be more intuitive.
I was already surprised to learn that "wrench menu settings" can be for the
current tab only. I had erroneously assumed it applies for all tabs (even
windows) from "then on". Therefore, to make a "wrench menu setting" (in this
case "RTS") even more fine grained may be unintuitive for some users.

[sschmitz, 2012/08/03 23:25:26]

Later: After more talk with Dan, I changed it back to RTS Per Entry, i.e. not
for past entries in the current tab, but only the current and further new ones.

+  } else {
+    entry->SetIsOverridingUserAgent(true);
+    current_tab->SetUserAgentOverride(
+        webkit_glue::BuildUserAgentOverrideForTabletSiteFromUserAgent(
+            content::GetUserAgent(entry->GetOriginalRequestURL())));
+  }
+  current_tab->GetController().Reload(true);
+}
+
 void ToggleFullscreenMode(Browser* browser) {
   browser->fullscreen_controller()->ToggleFullscreenMode();
 }