learn
-----

With no arguments, all of the flags that control chunking are displayed. If
arguments are provided, they will alter the learning environment as described
below.  Many of these values can also be controlled via watch (see Section
See watch on page See watch).

SYNTAX

 (learn arg*)
        arg  ::=  on | off
        arg  ::=  all-goals | bottom-up
        arg  ::=  noprint | print | full-print
        arg  ::=  notrace | trace | full-trace


 Below is a list of argument pairs; the first one is the default.
     
     
   * on/off
     On turns chunking on and off 
     turns it off. 
     If learning is set to off, it can be turned on and off at
     any point during a run.
     
   * all-goals/bottom-up
      The all-goals setting builds
     productions whenever a subgoal returns a result.  The alternate setting,
     bottom-up will cause productions to be built only for subgoals that
     have not yet had any subgoals with chunks built.
     
   * print/noprint/full-print
     With print,
     production names are printed as they are created.  With noprint,
     nothing is printed.  With full-print, the full production is printed
     when it is created.
     
   * trace/notrace/full-trace
     With trace, every
     chunk that has settings of :firings-wmes and {:firings-perferences} is traced.  With full-trace, the building of the production
     is also traced.



EXAMPLE

Soar> learn on full-print