Prototyping the infrastructure and UI together

[Michael Allan]

(following on today's skype talk between Alex and I)

Alex is thinking of developing a UI over the existing infrastructure.
Architecturally it would fall in the same category as crossforum
theatre (thick integrational client).  It's a category that's
invisible to the infrastructure (as clients usually are), so the
infrastructure may support any number of different types and instances
of these UIs.  (David's proposal might also fall in this category.)

      user interface (UI)

             |   visibility
             V

       infrastructure

The infrastructure consists of what we've already prototyped in
running code including the pollwiki, pollserver, and difference
bridge, plus other bits that are still only planned.  Its design is
described at an abstract level in a scattered set of architectural
docs.  The docs have no proper overview yet, but here are some
indeces:

  * Voting media: http://zelea.com/project/autonomy/a/vot/vot.xht
  * Drafting media: http://zelea.com/project/autonomy/a/txt/txt.xht
  * Discussion media: http://zelea.com/project/autonomy/a/yak/yak.xht

Alex and I figured that a good way to move forward would be to:

  1. Read those docs to get an understanding of what's in the
     infrastructure or planned for it.  (I can help if you get stuck
     ofc, because some of the docs are very rough.)

  2. Sketch the design of the UI on a napkin, for example like this
     sketch for crossforum theatre:
     http://zelea.com/project/votorola/a/crossforum/theatrePlan.png

  3. Negotiate the required service interfaces in support of the UI,
     or the required design changes to the UI for any functions that
     cannot be serviced.

  4. Prototype the implementations on both sides, in tandem, while
     simultaneously firming up the architectual documents.

As long as the negotiations pan out and the development progresses on
the UI side, I can probably keep pace on the infrastructural side and
code whatever's needed for Alex's client, and maybe one or two others
(if anyone else wants to float one).  Just to be sure, we agreed that
it's all prototype work with no immediate assurances of production
quality or scalability.

(I'll reply to the other posts soon)

Mike