Profiles: Add a size check to ProfileImpl to make sure it no longer grows.

chrome/browser/profiles/profile_impl.cc

Index: chrome/browser/profiles/profile_impl.cc
diff --git a/chrome/browser/profiles/profile_impl.cc b/chrome/browser/profiles/profile_impl.cc
index cdef23c94bb42390a98b6e3f6b8d2928c1b787bb..011a514c74caac54467ba49bd055ef7fda41b135 100644
--- a/chrome/browser/profiles/profile_impl.cc
+++ b/chrome/browser/profiles/profile_impl.cc
@@ -149,6 +149,22 @@ using content::UserMetricsAction;
 
 namespace {
 
+// Constrict us to a very specific platform and architecture to make sure
+// ifdefs don't cause problems with the check.
+#if defined(OS_LINUX) && defined(TOOLKIT_GTK) && defined(ARCH_CPU_X86_64)
+// Make sure that the ProfileImpl doesn't grow. We're currently trying to drive
+// the number of services that are included in ProfileImpl (instead of using
+// ProfileKeyedServiceFactory) to zero.
+//
+// If you don't know about this effort, please read:
+//   https://sites.google.com/a/chromium.org/dev/developers/design-documents/profile-architecture
+//
+// REVIEWERS: Do not let anyone increment this. We need to drive the number of
+// raw accessed services down to zero. DO NOT LET PEOPLE REGRESS THIS UNLESS
+// THE PATCH ITSELF IS MAKING PROGRESS ON PKSF REFACTORING.
+COMPILE_ASSERT(sizeof(ProfileImpl) == 656u, profile_impl_size_unexpected);

[Miranda Callahan, 2012/01/06 15:34:57]

Should we instead assert sizeof(ProfileImpl) <= 656u, so that we're allowed to
remove services without triggering errors?

[Elliot Glaysher, 2012/01/06 18:23:58]

I'm not crazy about this, but I changed it.

Using <= means I'll have to constantly maintain this, or else someone will
remove an object and then someone else may add an object without this
triggering.

+#endif
+
 // Delay, in milliseconds, before we explicitly create the SessionService.
 static const int kCreateSessionServiceDelayMS = 500;