Bundle fake Widevine component manifest for stub CDM

chrome/chrome_dll_bundle.gypi

Index: chrome/chrome_dll_bundle.gypi
diff --git a/chrome/chrome_dll_bundle.gypi b/chrome/chrome_dll_bundle.gypi
index 5b49c215afaa1659644dae695673cbc7356dd7df..40f0f311be0c372a00c352e9fbcb962e68e3a0f9 100644
--- a/chrome/chrome_dll_bundle.gypi
+++ b/chrome/chrome_dll_bundle.gypi
@@ -136,27 +136,19 @@
     },
     {
       # The adapter is not a complete library on its own. It needs the Widevine
-      # CDM to work.
+      # CDM to work. The CDM and manifest are either the actual Widevine CDM
+      # and manifest or stubs used for testing only. The choice is made within
+      # the corresponding Widevine targets based on branding.
       'destination': '<(PRODUCT_DIR)/$(CONTENTS_FOLDER_PATH)/Libraries/<(widevine_cdm_path)',
-      'files': [],
-      'conditions': [
-        ['branding == "Chrome"', {
-          'files': [
-            '<(PRODUCT_DIR)/<(widevine_cdm_path)/libwidevinecdm.dylib',
-            '<(PRODUCT_DIR)/<(widevine_cdm_path)/widevinecdmadapter.plugin',
-          ],
-        }],
+      'files': [

[ddorwin, 2016/07/18 18:35:17]

Don't we still need a condition for Chrome || enable_widevine?

[wdzierzanowski, 2016/07/19 11:35:17]

Ah, of course. Thanks for catching it!

[wdzierzanowski, 2016/07/19 11:35:17]

Done.

+        '<(PRODUCT_DIR)/<(widevine_cdm_path)/libwidevinecdm.dylib',

[ddorwin, 2016/07/18 18:35:17]

It's not clear to me that putting the stub CDM or manifest in the bundle is the
correct thing to do since this will actually be in the app bundle. This works
fine for testing but you wouldn't want to ship it.

[wdzierzanowski, 2016/07/19 11:35:17]

You're right, this must not be shipped.

I'm a bit on the fence about it myself. On the one hand, Opera doesn't really
need this change. The target that builds the encrypted media browser tests
doesn't even use this and we need to take care of bundling Widevine for tests
separately anyway. I don't know about the other browsers.

On the other hand, adding the bundling here for enable_widevine==1 has the
advantage of making this and similar changes testable in a Chromium-branded
build. Currently, it's only testable in an official Chrome build AFAIK.

Given that we already have to make sure we don't ship the stub CDM that is
already put in the bundle by Opera-specific build files to enable its use in
tests, I'm leaning towards the latter though.

[wdzierzanowski, 2016/07/21 08:47:06]

I feel like I need to clarify some bits of what I wrote:
Here, by "this change" I actually meant just the changes to the build files in
chrome/.
 
This time around, "this change" refers to the entire CL.

Sorry for the confusing language :)

+        '<(PRODUCT_DIR)/<(widevine_cdm_path)/widevinecdmadapter.plugin',
       ],
     },
     {
       'destination': '<(PRODUCT_DIR)/$(CONTENTS_FOLDER_PATH)/Libraries/WidevineCdm',
-      'files': [],
-      'conditions': [
-        ['branding == "Chrome"', {
-          'files': [
-            '<(PRODUCT_DIR)/WidevineCdm/manifest.json',
-          ],
-        }],
+      'files': [
+        '<(PRODUCT_DIR)/WidevineCdm/manifest.json',
       ],
     },
     {