Remove command line/field trial support and configs for Isolate Extensions.

chrome/browser/site_details.cc

 // Copyright 2013 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "chrome/browser/site_details.h"
 
 #include "base/metrics/histogram_macros.h"
 #include "content/public/browser/browser_thread.h"
 #include "content/public/browser/render_frame_host.h"
 #include "content/public/browser/render_process_host.h"
@@ skipping 22 matching lines @@
     case ISOLATE_NOTHING:
       return false;
     case ISOLATE_ALL_SITES:
       return true;
     case ISOLATE_HTTPS_SITES:
       // Note: For estimation purposes "isolate https sites" is really
       // implemented as "isolate non-http sites". This means that, for example,
       // the New Tab Page gets counted as two processes under this policy, and
       // extensions are isolated as well.
       return !site.SchemeIs(url::kHttpScheme);
-    case ISOLATE_EXTENSIONS: {

[ncarter (slow), 2017/05/01 20:15:19]

I don't think it's necessary to remove this?  If anything we should remove
ISOLATE_NOTHING and keep ISOLATE_EXTENSIONS, since it's the new baseline.

[nasko, 2017/05/01 21:25:37]

The way I read this code is "What should our numbers look if we assume
non-default process model?". Since the default process model is now to isolate
extensions, it is definitely the new baseline and I assumed ISOLATE_NOTHING to
be the value indicating baseline. We were already isolating WebUI, actual
extensions background pages, so ISOLATE_NOTHING wasn't really a great name in
the past either. Would changing it to ISOLATE_DEFAULT be a better naming?

Irrespective of the naming, would we need this code below? It seems the default
process model already does it, so calculating expected values doesn't seem
necessary, when they are the actual values.

[ncarter (slow), 2017/05/01 21:37:43]

The value of computing the simulated ISOLATE_EXTENSIONS is mainly in being able
to cross validate the simulation (which is imperfect) with the actual.

It's alsoe potentially useful to compare ISOLATE_HTTPS_SITES to the simulated
ISOLATE_EXTENSIONS metric, since the NoLimit/LowerBound/Estimate/CountEstimate
metrics are computed the same way, and are directly comparable. Converting from
the actual process count is requires quite a few more assumptions.

-#if !BUILDFLAG(ENABLE_EXTENSIONS)
-      return false;
-#else
-      if (!site.SchemeIs(extensions::kExtensionScheme))
-        return false;
-      extensions::ExtensionRegistry* registry =
-          extensions::ExtensionRegistry::Get(browser_context);
-      const extensions::Extension* extension =
-          registry->enabled_extensions().GetExtensionOrAppByURL(site);
-      return extension && !extension->is_hosted_app();
-#endif
-    }
   }
   NOTREACHED();
   return true;
 }
 
 content::SiteInstance* DeterminePrimarySiteInstance(
     content::SiteInstance* site_instance,
     SiteData* site_data) {
   // Find the BrowsingInstance this WebContents belongs to by iterating over
   // the "primary" SiteInstances of each BrowsingInstance we've seen so far.
@@ skipping 204 matching lines @@
                            num_isolated_site_instances[ISOLATE_HTTPS_SITES]);
   UMA_HISTOGRAM_COUNTS_100(
       "SiteIsolation.IsolateHttpsSitesProcessCountLowerBound",
       process_count_lower_bound[ISOLATE_HTTPS_SITES]);
   UMA_HISTOGRAM_COUNTS_100(
       "SiteIsolation.IsolateHttpsSitesProcessCountEstimate",
       process_count_estimate[ISOLATE_HTTPS_SITES]);
   UMA_HISTOGRAM_COUNTS_100(
       "SiteIsolation.IsolateHttpsSitesTotalProcessCountEstimate",
       process_count_estimate[ISOLATE_HTTPS_SITES] + non_renderer_process_count);
-
-  // ISOLATE_EXTENSIONS metrics.
-  UMA_HISTOGRAM_COUNTS_100("SiteIsolation.IsolateExtensionsProcessCountNoLimit",
-                           num_isolated_site_instances[ISOLATE_EXTENSIONS]);
-  UMA_HISTOGRAM_COUNTS_100(
-      "SiteIsolation.IsolateExtensionsProcessCountLowerBound",
-      process_count_lower_bound[ISOLATE_EXTENSIONS]);
-  UMA_HISTOGRAM_COUNTS_100(
-      "SiteIsolation.IsolateExtensionsProcessCountEstimate",
-      process_count_estimate[ISOLATE_EXTENSIONS]);
-  UMA_HISTOGRAM_COUNTS_100(
-      "SiteIsolation.IsolateExtensionsTotalProcessCountEstimate",
-      process_count_estimate[ISOLATE_EXTENSIONS] + non_renderer_process_count);
 }