Adding pagesets for gmail and plus. Creating endure benchmark

tools/perf/page_sets/gmail_expand_collapse_conversation.json

Index: tools/perf/page_sets/gmail_expand_collapse_conversation.json
diff --git a/tools/perf/page_sets/gmail_expand_collapse_conversation.json b/tools/perf/page_sets/gmail_expand_collapse_conversation.json
new file mode 100644
index 0000000000000000000000000000000000000000..eda38e74afd80df58fb44cb1469c3e4028b4fdce
--- /dev/null
+++ b/tools/perf/page_sets/gmail_expand_collapse_conversation.json
@@ -0,0 +1,26 @@
+{
+  "description": "Chrome Endure test for GMail.",
+  "archive_data_file": "../data/gmail_expand_collapse_conversation.json",
+  "credentials_path": "../data/credentials.json",
+  "user_agent_type": "desktop",
+  "pages": [
+    {
+      "url": "https://mail.google.com/mail/u/0/#inbox/13c6a141fa95ffe0",

[edmundyan, 2013/08/26 21:31:55]

How do we ensure we enter a long conversation mail?

[dennis_jeffrey, 2013/08/27 20:02:35]

Right now we have to ensure the account is set up in such a way that a long
conversation e-mails opens up (and the e-mails in the account don't change from
run to run).  WebPageReplay can help when we're replaying things, but in
recording mode, we still have to ensure a long conversation e-mail is used.  In
the future we should think about how to better ensure this.

[edmundyan, 2013/08/27 23:01:48]

So is the explicit URL acceptable?  This won't work for other accounts.  The old
chrome endure test just clicked the 1st email in the inbox and hoped it would
have replies.

[dennis_jeffrey, 2013/08/27 23:17:22]

I'd say it's acceptable with a TODO added to the code that we should implement a
better approach.

+      "why": "Expand and Collapse a long conversation",
+      "credentials": "google",
+      "navigate_steps": [
+        { "action": "navigate" },
+        { "action": "endure" },
+        { "action": "wait", "condition": "duration", "seconds": 10 }

[edmundyan, 2013/08/26 21:31:55]

Need to ensure that all expanded emails have their contents downloaded (this is
more for WPR to correctly record).  There should be a more precise way to know
when all downloading is complete.

+      ],
+      "endure": [
+        { "action": "wait", "condition": "element", "selector": "img[alt=\"Expand all\"]" },
+        { "action": "click_element", "selector": "img[alt=\"Expand all\"]" },
+        { "action": "wait", "condition": "duration", "seconds": 1 },
+        { "action": "wait", "condition": "element", "selector": "img[alt=\"Collapse all\"]" },
+        { "action": "click_element", "selector": "img[alt=\"Collapse all\"]" },
+        { "action": "wait", "condition": "duration", "seconds": 1 }
+      ]
+    }
+  ]
+}