Add a SetCanonicalCookie method for CookieMonster.

net/cookies/cookie_monster_unittest.cc

Index: net/cookies/cookie_monster_unittest.cc
diff --git a/net/cookies/cookie_monster_unittest.cc b/net/cookies/cookie_monster_unittest.cc
index 0b60812a4415f788f663da2d745ee87318b4dd2b..a29b555f534e2195e0e70070d49d26b7e94fcec5 100644
--- a/net/cookies/cookie_monster_unittest.cc
+++ b/net/cookies/cookie_monster_unittest.cc
@@ -148,6 +148,20 @@ class CookieMonsterTestBase : public CookieStoreTest<T> {
     return callback.result();
   }
 
+  bool SetCanonicalCookie(CookieMonster* cm,
+                          CanonicalCookie cookie,
+                          bool secure_source,
+                          bool modify_http_only) {
+    DCHECK(cm);
+    ResultSavingCookieCallback<bool> callback;
+    cm->SetCanonicalCookieAsync(
+        cookie, secure_source, modify_http_only,
+        base::Bind(&ResultSavingCookieCallback<bool>::Run,
+                   base::Unretained(&callback)));
+    callback.WaitUntilDone();
+    return callback.result();
+  }
+
   int DeleteAllCreatedBetween(CookieMonster* cm,
                               const base::Time& delete_begin,
                               const base::Time& delete_end) {
@@ -2303,7 +2317,7 @@ TEST_F(CookieMonsterTest, WhileLoadingDeleteAllGetForURL) {
   // When passed to the CookieMonster, it takes ownership of the pointed to
   // cookies.
   cookies.push_back(
-      CanonicalCookie::Create(kUrl, "a=b", base::Time(), CookieOptions()));
+      CanonicalCookie::Create(kUrl, "a=b", base::Time::Now(), CookieOptions()));

[mmenke, 2017/06/16 15:56:28]

Why is this change needed in this CL?

[Randy Smith (Not in Mondays), 2017/06/16 21:38:00]

Because I added a DCHECK() in CanonicalCookie::Create() that creation time was
not null.  That combination of changes isn't strictly needed in this CL, but I
was refining the specification of creation time (because I wanted to allow
consumers to pass in a null creation time and have it set on insertion for the
new method so that the new method didn't result in duplicate creation time in
the backing store) and I took the attitude that if a routine specifically took
creation time as one of a small number of arguments, the caller should actually
set it.  I can pull this change (or split it out into a separate CL, but it's
pretty small) if you want, but I called out somewhere else that CC::Create()
does depend on a non-null creation time, so I think there's some value in
forcing it to be non null by spec.

This, obviously, doesn't do jack for consumers manually setting the creation
time on two cookies to be the same; that'll mess up the store.  Existing
problem, but one that continues to bother me :-{.

   ASSERT_TRUE(cookies[0]);
   store->commands()[0].loaded_callback.Run(std::move(cookies));