Welcome everyone,

sorry that it took a while. I was busy with other important development.

I decided to change a little detail in the thread model again. The 
original plan was to only have a worker thread for a remote node if the 
local node is subscribed to any set originating from that remote node 
and a default worker for all other nodes (there could only be config 
messages from there that are important.

The potential danger here is, that subscribing causes the internal 
message destination from the default worker to change to a new thread, 
the node specific worker. Coordinating that to prevent message loss is 
not trivial.

The win of having a default worker instead of blindly starting one for 
every node out there is in my favorite reference scenario (one master, 
two slaves, one cascaded slave) one or two less threads on every node. 
Considering that we are talking about 8-10 threads here anyway, and that 
this only means that one thread will do the same work otherwise 3 
threads would do, I don't see a big win here.

Documentation update will follow.

The main engine code is advancing. So far I have threads that are 
generating SYNC events and cleanup, and the local and remote event 
listener threads. The next steps will be creating the remote worker 
thread that processes and forwards event information (and will some day 
in the future also replicate some data).


Jan

-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck at Yahoo.com #