Reloading page after installing app should bring it into correct process.

content/browser/tab_contents/navigation_controller_impl.cc

Index: content/browser/tab_contents/navigation_controller_impl.cc
diff --git a/content/browser/tab_contents/navigation_controller_impl.cc b/content/browser/tab_contents/navigation_controller_impl.cc
index 3c5ddeee33263d3ea536685549e4fd44ff255eb6..9c0ca52aa0bcf1230f90a2c66966bd617bc673fe 100644
--- a/content/browser/tab_contents/navigation_controller_impl.cc
+++ b/content/browser/tab_contents/navigation_controller_impl.cc
@@ -265,9 +265,29 @@ void NavigationControllerImpl::ReloadInternal(bool check_for_repost,
   } else {
     DiscardNonCommittedEntriesInternal();
 
-    pending_entry_index_ = current_index;
-    entries_[pending_entry_index_]->SetTransitionType(
-        content::PAGE_TRANSITION_RELOAD);
+    NavigationEntryImpl* entry = entries_[current_index].get();

[Charlie Reis, 2012/01/26 01:20:23]

GetEntryAtIndex

[nasko, 2012/01/26 20:25:57]

I've used direct access to the collection, since I get back a
NavigationEntryImpl*, whereas this call returns NavigationEntry*. Since site
instance is not publicly available through NavigationEntry*, I would need to
cast it. This version seems more intuitive to get the proper pointer type back.

[Charlie Reis, 2012/01/26 21:49:45]

Ah.  Yes, that's fine then.  I've emailed John about adding SiteInstance to the
public NavigationEntry interface, but that will be for a separate CL.

+    SiteInstance* site_instance = entry->site_instance();
+
+    // If we are reloading an entry that no longer belongs to the current
+    // site instance (for example, refreshing a page for just installed app),
+    // the reload must happen in a new process and behave as new navigation

[Charlie Reis, 2012/01/26 01:20:23]

Nit: let's split out the "behave as new navigation" into a separate sentence
that says a bit more.  It's important to note that we need a new entry with a
different SiteInstance and fresh page_id, so we treat it as a new navigation
(which happens to clear the forward history).

[nasko, 2012/01/26 20:25:57]

Done.

+    // when it comes to history.
+    if (site_instance != NULL &&

[Charlie Reis, 2012/01/26 01:20:23]

More common to drop the != NULL.  And actually, should we be asserting that it
isn't NULL?  If we're reloading an entry, I would never expect it to have a NULL
SiteInstance.

[nasko, 2012/01/26 20:25:57]

Done.

+        site_instance->HasWrongProcessForURL(entry->GetURL())) {
+      // Create a navigation entry that resembles the current one

[Charlie Reis, 2012/01/26 01:20:23]

Nit: perhaps mention which things must not be copied (page ID, SiteInstance, and
content_state), and end comments with a period.

[nasko, 2012/01/26 20:25:57]

Done.

+      NavigationEntryImpl* nav_entry = NavigationEntryImpl::FromNavigationEntry(
+          CreateNavigationEntry(
+              entry->GetURL(), entry->GetReferrer(), entry->GetTransitionType(),
+              false, entry->extra_headers(), browser_context_));
+
+      nav_entry->set_is_cross_site_reload(true);
+      pending_entry_ = nav_entry;
+    } else {
+      pending_entry_index_ = current_index;
+      entries_[pending_entry_index_]->SetTransitionType(
+          content::PAGE_TRANSITION_RELOAD);
+    }
+
     NavigateToPendingEntry(reload_type);
   }
 }
@@ -582,6 +602,13 @@ bool NavigationControllerImpl::RendererDidNavigate(
   // assigned one in NavigateToPendingEntry.
   DCHECK(pending_entry_index_ == -1 || pending_entry_->site_instance());
 
+  // If we are doing a cross-site reload, we need to replace the existing
+  // navigation entry, not add another entry to the history. This has the side
+  // effect of removing forward browsing history, if such existed.
+  if (pending_entry_ != NULL) {
+    details->did_replace_entry =  pending_entry_->is_cross_site_reload();

[Charlie Reis, 2012/01/26 01:20:23]

Nit: extra space.

+  }
+
   // is_in_page must be computed before the entry gets committed.
   details->is_in_page = IsURLInPageNavigation(params.url);