encryptMessage is now called in a loop, because problems may happen by steps (e.g. trust issue after getting missing bundles).
Fixed a mix between "if" and "elif" in problems handling,
state handling is not using mmap anymore but an unix socket instead.
State change can now be managed.
Bridge signals are disabled when frontend is paused.
- public bundles are now retrieved only when needed (i.e. on MissingBundleException when encrypting a message)
- a feedback is given to user when a message is received but it has not been encrypted for our device
- fixed feedback on decryption error if message come from an other device of the client (i.e. it's a carbon copy)
- setBundle if called in a "finally" if republish_bundle is set after decrypting a message
- carbon prioriy must be bigger than ones of encryption plugins, because the carbon copied messages must be unwrapped to be decrypted, this patch fixes it
- "sent" carbon elements was only stored and notified to frontends, it is now unwrapped and injected in normal traffic, this way other plugin can manage it, in particular encrypted elements can be decrypted
jid caching in Twisted use a dict (__internJIDs) which associate a jid as string to its jid.JID instance.
But this caching doesn't handle mutability of the objects, resulting wrong jid returned when a modification is done (e.g. if resource is changed).
This is a temporary workaround until the patch is proposed and merged upstream.
note: this is a partial fix because the caching is also leaking memory (cf. https://twistedmatrix.com/trac/ticket/5122)
So far, and for historical reasons, blog items data where serialised using a unicode: unicode dict, which was causing trouble for many types of values (timestamps, booleans, lists).
This patch changes it by serialising the whole items before going to bridge, and deserialising it when going back. This way, complex data can be used easily in items.
This impact psEvent and serItemsData* methods which are renamed transItemsData* because there are not always serialising anymore (a new argument "serialise" allows to specify it).
When editing a blog post in jp, metadata are now more easy to manipulate, specially lists like tags.
indexes were missing on foreign keys for "message", "subject" and "thread" tables, seriously impacting performances.
In addition to those indexes, two indexes have been added to speed ordering by timestamp on "history", and one on "files" table.
history.rowid is not used anymore as there is an index on (profile_id, received_timestamp) which will speed up the query.
Primary keys order has been changed to use automatic index in most common cases (filtering by profile_id then namespace).
