-
Committer:
Ken VanDine
-
Date:
2012-11-06 21:22:48 UTC
-
Revision ID:
ken.vandine@canonical.com-20121106212248-r2ijp8f386o8pxsx
Don't make the model leader writable only, that doesn't work out well with
multiple processes that need to connect to the model. They all try to keep
the model alive, then stop getting synchronized.
Attempt to flush_revision_queue in persist_model, but silently ignore
failures. The flush could only fail if the Dee.SharedModel hasn't been
created yet, at which point we don't care.