Friday, January 13, 2012

A demonstable prototype: what is left to do?

Clique Space(TM) is almost at a demonstrable state; at least this is what I feel. The Agent Collaboration's pulse message is something that needs to be done to get it to this state.

The Agent Devices still lack what is necessary to collaborate. The pulse message is still swimming around my head as a set of ideas that have not found a place to roost. As much as can be told, engaged Agent Devices send pulses to each other. These pulses will either alert collections of collaborating Agent Devices to the change in a Clique's state (additions or deletions of Elements) or changes in an Element's state (additions or deletions of Limiting Constraints). My most recent SVN revision committed a start on the pulse message, but still, some thought will have to be applied to the specific utility of the variations of pulse message that will be necessary. This will take some time.

My current concern (I'm hoping one revision will solve it) is that I want to move the declaration of the undisclosed Element from the Client Device project to the shared Clique Space project so the Agent Device can capitalise on the use of the undisclosed Element.

Everything else described below has to do with the concept's implementation once the proof-of-concept has been achieved.

Late 2010, I put a blog entry together like this one estimating how far I thought I had to go. I actually don't think I am much further along now than what I said in that entry, but casting my mind back to that time, I can say that my progress from that point in time has been significant. The current system is certainly next to useless in terms of the fact that Media Profile customisations are necessary to make some of the more popular media Clique Space aware. Nothing has changed currently.

Certainly, the implementation is still ugly. The administrator Client Device still doesn't yet have - because I haven't prioritised time to put in - a nice graphical user interface, although I feel significantly more comfortable that the console output it is giving me now makes more sense than it did back in October 2010. My greatest feeling of progress from 2010 has been in the degree to which forming, joining, leaving and disbanding Cliques have increased in their flexibility.

Agent Devices are still fragile and inflexible; they lack even rudimentary logic necessary to maintain a stable Clique Space in environmentally adverse conditions (mainly because they lack an implementation of the pulse message), and I have still to host different Agent Devices on different machines although I have delayed doing this because of the fact that I believe it is a trivial consideration. Progress has perhaps been made on memory performance and hygiene, and some concurrency control has been implemented although anything approaching true transaction logic is perhaps still a while off.

I am finding that when implementing such things as the Agent Device's engage/disengage semantics, it at least appears that concerns around security are being watered down by mechanisms that have been necessary for implementation.


No comments:

Post a Comment