[Extensions Sync] Don't apply sync data for unsyncable extensions

chrome/browser/extensions/chrome_test_extension_loader.cc

Index: chrome/browser/extensions/chrome_test_extension_loader.cc
diff --git a/chrome/browser/extensions/chrome_test_extension_loader.cc b/chrome/browser/extensions/chrome_test_extension_loader.cc
index 9bc6272ed4406d049d10d04c4aaf9a6866a45551..aef672fd5907fcd6fa596c0eca47eef021934328 100644
--- a/chrome/browser/extensions/chrome_test_extension_loader.cc
+++ b/chrome/browser/extensions/chrome_test_extension_loader.cc
@@ -104,6 +104,10 @@ base::FilePath ChromeTestExtensionLoader::PackExtension(
     return base::FilePath();
   }
 
+  if (!temp_dir_.CreateUniqueTempDir()) {

[karandeepb, 2017/06/09 22:53:36]

How did this work before w/o this?

[Devlin, 2017/06/13 16:14:44]

It didn't - turns out, we never exercised this code.

+    ADD_FAILURE() << "Could not create unique temp dir.";
+    return base::FilePath();
+  }
   base::FilePath crx_path = temp_dir_.GetPath().AppendASCII("temp.crx");
   if (base::PathExists(crx_path)) {
     ADD_FAILURE() << "Crx path exists: " << crx_path.value()
@@ -118,7 +122,8 @@ base::FilePath ChromeTestExtensionLoader::PackExtension(
     return base::FilePath();
   }
 
-  base::FilePath* pem_path_to_use = &fallback_pem_path;

[karandeepb, 2017/06/09 22:53:36]

Isn't this also correct? This change is just for more clarity? 

[Devlin, 2017/06/13 16:14:44]

It's actually not correct - if the path_to_use isn't empty, the extension
creator looks for it (and in this case, we assert it doesn't exist).

+  base::FilePath empty_path;
+  base::FilePath* pem_path_to_use = &empty_path;
   if (!pem_path_.empty()) {
     pem_path_to_use = &pem_path_;
     if (!base::PathExists(pem_path_)) {