[fsp] Add support for creating directories.

chrome/common/extensions/api/file_system_provider.idl

Index: chrome/common/extensions/api/file_system_provider.idl
diff --git a/chrome/common/extensions/api/file_system_provider.idl b/chrome/common/extensions/api/file_system_provider.idl
index 471c6a1ff7d44bb0ca8e694a0680fce31f1c92c5..c08a884751c23d27f2353b529b76c2c7210c091b 100644
--- a/chrome/common/extensions/api/file_system_provider.idl
+++ b/chrome/common/extensions/api/file_system_provider.idl
@@ -109,6 +109,15 @@ namespace fileSystemProvider {
     double length;
   };
 
+  // Options for the <code>onCreateDirectoryRequested()</code> event.
+  dictionary CreateDirectoryRequestedOptions {
+    DOMString fileSystemId;
+    long requestId;
+    DOMString directoryPath;
+    boolean exclusive;
+    boolean recursive;
+  };
+
   // Callback to receive the result of mount() function.
   callback MountCallback = void([nodoc, instanceOf=DOMError] object error);
 
@@ -218,6 +227,15 @@ namespace fileSystemProvider {
         ReadFileRequestedOptions options,
         FileDataCallback successCallback,
         ProviderErrorCallback errorCallback);
+
+    // Raised when creating a directory is requested. If <code>exclusive</code>
+    // is set to true, then the operation should fail if the target directory
+    // already exists. If <code>recursive</code> is true, then all of the

[not at google - send to devlin, 2014/07/08 16:29:43]

it would be nice (generally speaking) if comments like "If
<code>exclusive</code> is set to true, then the operation should fail if the
target directory already exists" were actually on the
CreateDirectoryRequestedOptions dictionary itself, rather than here.

[mtomasz, 2014/07/09 02:08:30]

Let me clean up all of the comments in a separate CL.

+    // missing directories on the directory path should be created.
+    [maxListeners=1, nodoc] static void onCreateDirectoryRequested(

[not at google - send to devlin, 2014/07/08 16:29:43]

why is this nodoc?

[mtomasz, 2014/07/09 02:08:30]

We don't want it on developer.chrome.com yet.

[not at google - send to devlin, 2014/07/09 02:34:39]

why not?

[not at google - send to devlin, 2014/07/09 02:39:10]

i.e. because if it's that the API isn't ready then a better way to do all of
this is to restrict it to dev channel or whatever rather than not documenting
it.

[mtomasz, 2014/07/09 04:03:11]

The API is already restricted to build from sources only. However, since the
read-only version is almost finished, we may want to enable it soon for dev
users. However, R/W is far from ready, and I don't want users to use it yet.

Hence, my plan was to avoid publishing documentation for R/W functions until it
is reasonably working.

However, I don't have very strong feelings here. If you think that we should
publish the doc for stuff being worked on, I'll remove [nodoc]. Let me know.

[not at google - send to devlin, 2014/07/09 13:28:58]

Ah I see. Yeah, nodoc is fine then.

+        CreateDirectoryRequestedOptions options,
+        ProviderSuccessCallback successCallback,
+        ProviderErrorCallback errorCallback);
   };
 };