Separate invalidator and sync client ID (part 1/2)

sync/protocol/sync.proto

Index: sync/protocol/sync.proto
diff --git a/sync/protocol/sync.proto b/sync/protocol/sync.proto
index 44dd59086e16e83dc32fbb47080a9c86ee163581..43a1439d10842e93dd594856b12e550b81862502 100644
--- a/sync/protocol/sync.proto
+++ b/sync/protocol/sync.proto
@@ -565,6 +565,11 @@ message ClientToServerMessage {
   // The client should set this on every message sent to the server, though its
   // member fields may often be unset.
   optional ClientStatus client_status = 13;
+
+  // The ID that our invalidation client used to identify itself to the server.
+  // Sending the ID here allows the server to not send notifications of our own
+  // changes to our invalidator.
+  optional string invalidator_client_id = 14;

[Nicolas Zea, 2013/02/20 23:23:15]

should this be part of the ClientToServerMessage or the CommitMessage?

[rlarocque, 2013/02/20 23:53:32]

The server only needs access to it for commit messages, but there's no harm in
sending it up on every request.

IIRC, Raz had a slight preference for including it in all requests, but I forget
why that was.  Perhaps it had something to do with future-compatibility?

 };
 
 message CommitResponse {