
hg commands

show file changes between two revisions
hg status --rev 7:8

launch WindDiff to show two revisions
hg vdiff --rev 7:8 six_states.py

show commit messages for all commits.
hg log -T compact
-------------------------------------

*) from __future__ import unicode_literals
   BOMBS IN python 2.7 for things like:
   sout = s.rjust(lmaxL[j], ' ')

*) write routine "solve_sim_w_dp"... i.e. make model with transition probs
   and use value iteration

*) in policy diagram, need special symbol for actions that don't matter.
   (e.g. in Sutton 5x5 gridworld, jumps from (4,1) to (0,1) no matte what.

*) Replace state_values with state_value_coll

*) fix examples for move to IntroRL

*) add a min_alpha to Alpha object (i.e. decay to min_alpha instead of 0.0)
   (Same for Epsilon)

*) make generic hyperparameter object a single input to solvers.
   (i.e. reduce the number of inputs to each solver)

*) generate HTML output

*) add percent convergence number to running average

*) set policy of tuple-states by tuple position entry.
   e.g. if state=(p1, p2), init policy as (1,*)='UP'
        include NOT (1,*)='DOWN'

*) build graphic environment inspector
   - hover over state to see actions and transition probabilities
   - mark deterministic and stochastic states.
   - show arrows to next_states with probability

*) ??? in transition_coll.set_transition, as transitions are added, update terminal_set.
   (may make more sense to make CRC on last terminal_set calculation)

*) add state_previous data to episode_stats.

========================== DONE ==============================

*) need all examples subdirectories to be in PyPi install
   
*) need script to run all __name__=="__main__" scripts.

*) add car rental 3D plot to docs

*) change max_steps to include sys.maxsize initial value. (2147483647)

*) show '"string"' annotation in policy diagram

*) fix pickle of environment.


