Hominid Problem Active Classes

1
Hominid Problem - Active Classes
Ref Page-Jones FOOD Ch. 1, pp 1-8, 1.11(13),
1.12 p.56

• My proposed solution uses two Active Classes
• AC1 Hominid HM1 with state Direction d in
  E, N, W, S
• AC2 HominidLocation HL1 with state index
  i in 0..4
• The first AC has one Active Instance
• AI1 HD AI.state direction in enumE, N,
  W, S
• Enum order selected so TL is d (d)4 and
  TR is d ( d--)4
• The second AC has two active instances
• AI2 HX AI.state column index of its Grid
  Square.
• AI3 HY AI.state row index of its Grid
  Square

2
Hominid Supporting Classes

• There is one main program or singleton class
  PathAssigner (PA) with one method navigation.
• There are two passive classes SQuare and its
  GriD container class., with access methods.
• GD and SQ are constant, so they dont have
  behavioral states.
• HD has the fkey SQid (and SQid_pp)
• SQ has the fkey GDid (and GDid_pp)
• HD has the fkey PAid (so PA will have HDid_fcp)

3
Object Communication Diagram (OCD)
Illustrates Object Classes
PA HD HL
TL TR Adv
Incr Decr
4
UML Collaboration Diagram (UCD)
Illustrates Object Instances
PA HD1 HL1HX HL2 HY
TL TR Adv
Incr Decr
Incr Decr
5
UML Class or E-R Diagram
1..1
PathAssigner PA
1..1
Grid GD
1..1
1..1
0..
Square SQ
0..
Hominid HM
1..1
1..1
1. .1
2..2
HominidDirection HD
HominidLocation HL
6
UML Class or E-R Diagram
(Including Active Classes and Instances for
take-home exam)
AC
1..1
PathAssigner PA navigate()
3..3
1..1
Grid GD
AI state
1..1
0..
1..1
0..
Square SQ onPath()? x y
1..1
0..
0..1
Hominid HD
1..1
(Hominid HM merged into HominidDirection HD
HD.d AIid_pp-gtstate HL.i AIid_pp-gtstate)
2..2
HominidLocation HL
(Instance names HX, HY)
7
Pre-condition on Advance Cmd

• The particular algorithm which PA uses for
  advancing the hominid must insure that it will
  not attempt to advance into a wall.
• While debugging, we can assert or check this
  pre-condition for each Advance command that
  reaches the HD object from the (trusted) PA
  navigator
• assert( (HD.state E HX.i lt 4)
• (HD.state W HX.i gt0) (HD.state
  N HY.i gt 0)
• (HD.state S HY.i lt 4) )
• One sub-expression of this assert can be placed
  inside each case of switch(state).

8
Invariant on HD State Transition

• The particular algorithm which PA uses for
  advancing the hominid must also insure that it
  will not reverse direction.
• This condition is not a pre-condition but an
  invariant relationship between initial and final
  direction states for any TLTR sequence
• assert (((dOLD - dNEW)4 ) ! 2)
• Note the need to save the prior state dOLD while
  the next one dNEW is computed.

9
Sample Path for Test Case

• This task requires hand-tracing the algorithm in
  FOOD Ch. 1. As a test case, assume the following
  path which is square, with final cell equal to
  the starting one.

0,0
4,0
1,0
2,0 3,0
0,1
4,1
Exam2 Start
0,2
4,2
y
0,3
4,3
0,4
1,4
2,4
3,4
4,4
10
Behavioral Model

• The hominid occupies one cell and supports three
  action methods turnLeft, turnRight,
  advanceOneCell, and a Boolean query
  isFacingWall.
• It also has an initialization method insert
  (x,y,d).
• The PA.navigator method calls the HD state action
  method corresponding to the command or event
  type.
• The HD.Advance action calls the HL.Up or HL.Down
  method of the HX or Hy instance, depending on its
  direction state HN.d
• (2) Sequence Diagram (TBD)(3) EventType List
  (TBD)

11
Hominid Path Definition
x
0,0
4,0
1,0
2,0 3,0
0,1
4,1
0,2
4,2
STARTW
y
0,3
4,3
0,4
1,4
2,4
3,4
4,4
For Exam 2 I specified the starting state ltx,y,dgt
lt0,2,Wgt.
12
State Model for Hominid Location HL

• The x and y components of state have their own
  state diagrams, but they correspond to modulo-5
  up/down counter shown below. Only one counter can
  change at a time neither x nor y changes at the
  same time d does.
• The single HL class has a common name for the
  current state it inherits from its AC. I use the
  name i here
• HX.i AIcurr-gtstate, HY.i AIcurr-gtstate

Incr Incr Incr Incr
0
1
2
4
3
Decr Decr Decr Decr
13
State Model for Hominid Direction HD
Commands received from PA TR turnRight TL
turnLeft AD advance
Commands sent to HL Up increment Down
decrement
AD/HYUp
D N 1
TR
TR
TL
TL
D W 3
D E 0
TL
TL
TR
TR
AD/HXDown
D S 2
AD/HXUp
(Only ADV has actions other than the state
transition.)
AD/HYDown
14
(4A) State Action Code for HD

• include HDdefines.h
• // contains enums or defines for d e, N, W,
  S,
• //command TL, TR, ADV, and count Up,
  Down
• int HDaction(command)
• switch(state) // direction state of HD
• case E
• switch(command)
• case TL d (d)4 // next state
• case TR d (d-- )4 // next state
• case ADV HXcount(Up) // command to HX
• case N
• switch(command) // commands TL and TR as
  above
• case ADV(HY.count(Down)
• // states W and S as above
• return

15
(4B) State Action Code for HL

• include HDdefines.h / enum or defines for
  command /
• int HLcount(cmd)
• // no need to do switch(state) - action is
  independent of state HL.i
• // as long as pre-condition is met
• switch(cmd)
• case Up
• ifndef NDEBUG
• assert (HL.i lt 4)
• endif
• HL.i // here to save space
• case Down
• ifndef NDEBUG
• assert (HL.i gt 0)
• endif
• HL.i -- // case Down
• return
• // end count()