[CAP] CAP 1.2?

[Rex Buddenberg]

On Wed, 2007-10-24 at 10:22 -0400, Mick Jagger wrote:
> Hi,
> 	Just following up on a point Matt made regarding implementation.  The majority of problems that people have with CAP is understanding how to properly implement it.  A quick look through the list of publically available CAP feeds shows that most of them are publishing alerts that are invalid.
> 	I feel that there is a need for more documentation and development resources rather than a major redesign of the standard.  That should be the focus of any work at this point, instead of redesigning something that many people aren't using properly in the first place.
> 	Things like an FAQ, a developer's guide, a list of example alerts, etc.  I'd be willing to work on the FAQ and if people have questions, send them to the list and they can be compiled into something on the CAP cookbook.
> 

How the other standards bodies do this...

Typically a standard has knobs and switches written into it that
ultimately nobody uses.  Even the best of the 'minimal/extensible' type
have this happen to them.  There are two well-known ways to swab up this
post-standard clutter.

1.  Internet Engineering Task Force tries to prevent the clutter in the
first place.  By demanding that implementations and the standard (RFC)
proceed in parallel (e.g. reference implementations).  So the reality
check is built in.  While not foolproof, IETF's approach properly gets
high marks.

2.  IEEE 802 takes a slightly different approach which is also
effective.  In parallel to the standards body itself, there is an
industry association.  For example, 
  - IEEE 802.11 standard is matched by the WiFi industry association;
and 
  - IEEE 802.16 is matched by WiMAX. 
  - it's now history, but IEEE 802.3 was matched by ethernet in the same
fashion.
     The industry associations publish profiles which can be glibly
described as an instantiation of the standard with several of the
options eschewed -- a lowest common denominator.  These profiles are
routinely accompanied by test scripts or availability of a
neutral-ground interoperability testing lab.  

Both approaches occasionally use the bake-off technique which might make
sense with CAP.