Is DJ sound compared to Java?

1
Is DJ sound compared to Java?

• A comparison of effort and static analysis
  support when working with DJ and OO

2
Motivation

• DJ has the AI flavor that it automatically
  transforms programs when the context (class
  graph) changes.
• Some people see this power as a danger.
• We point out that when a proper software
  engineering process is followed there is no
  danger.
• Powerful tools can be used in bad ways.

3
Motivation

• Some people feel that DJ cannot give the same
  static error messages as OO tools. We show that
  this is not the case.
• Some people feel that DJ does not save any effort
  because the code needs to be tested after a class
  graph change. This is like saying high-level
  languages are bad. First address this point.

4
DJ programs are intentionally kept general
Class graph 1
DJ program (only methods)
Class graph 2
Class graph 3
5
DJ programs are intentionally kept general
Class graph 1 correct behavior
DJ program with constraints
Class graph 2 incorrect behavior
Class graph 3 rejected
6
DJ programs are intentionally kept general
Class graph 1 correct behavior
DJ program with constraints
Class graph 2 incorrect behavior
Distinction between and
requires testing
7
High-level control of oo program structure is
important

• From A to B selects paths with 50 edges
• From A via X to B selects path with 10 edges
• Small change big effect
• Once strategy correct, traversal guaranteed to be
  correct.
• Hand-coding of traversal is source of errors

8
What does DJ save during evolution? Terminology.

• Class graph G1 and traversal specification S1 for
  behavior B.
• Requirements change need to change G1 to G2. In
  general we might have to change traversal
  specification S1 to S2 to achieve desired
  behavior B. Note
• S1 is not necessarily a strategy it could be an
  English specification that says how to traverse
  G1-objects.

9
What does DJ save during evolution? Terminology.

• T(G,S) traversal methods for G and S.
• S is not necessarily a strategy it could be an
  English specification that says how to traverse
  G-objects.
• TG(G,S) traversal graph for G and S.
• Compatible(G,S) Is G compatible with S? For all
  edges (x,y) in S there is a path in G from x to y.

10
What does DJ save during evolution? Terminology.

• Diff(TG1, TG2), Diff(T(G1, S1), T(G2, S2)) list
  differences for each node.
• Errors(G2,T(G1,S1))) compiler error messages
• Errors(G2, G1) if G1 has an edge (A,B) then G2
  has such an edge, otherwise report error.

11
For AP library (used by DJ)

• Errors(G2, G1) if G1 has an edge (A,B) then G2
  has such an edge, otherwise report error.

12
What does DJ save during evolution?
Written by user
Inspected by user, generated by compiler

• Both Class graphs G1 and G2. Traversal methods
  T(G1,S1) and T(G2,S2) to be tested.
• DJ Strategy graphs S1, S2. Traversal graphs
  TG(G1,S1) simulating T(G1,S1) and TG(G2,S2)
  simulating T(G2,S2). Compatible(G2,S2).
  Errors(G2, Graph(TG(G1,S1))). Diff(TG(G1,S1), TG
  (G2,S2)).
• OO T(G1,S1) and T(G2,S2). Compatible(G2,T(G2,S2))
  . Errors (G2,T(G1,S1)). Diff(T(G1,S1), T(G2,S2)).

13
Comparing DJ versus OO during evolution effort
Written by user
Inspected by user, generated by compiler

• AP
• S1,TS ? S2
• Compatible(G2,S2)
• Errors(G2, Graph(TG(G1,S1)))
• Diff (TG(G1,S1), TG(G2,S2))

• OO
• T(G1,S1),TS ? T(G2,S2)
• Compatible(G2,T(G2,S2))
• Errors(G2,T(G1,S1))
• Diff (T(G1,S1), T(G2,S2))

Common G1 ? G2, test(T(G1,S1)), test(T(G2,S2)),
traversal change specification
TS Comparable effort testing DJ less effort
on average
14
Comparing DJ versus OO during evolution static
checking
Written by user
Inspected by user, generated by compiler

• AP
• S1,TS ? S2
• Compatible(G2,S2)
• Errors(G2, Graph(TG(G1,S1)))
• Diff (TG(G1,S1), TG(G2,S2))

• OO
• T(G1,S1),TS ? T(G2,S2)
• Compatible(G2,T(G2,S2))
• Errors(G2,T(G1,S1))
• Diff (T(G1,S1), T(G2,S2))

Common G1 ? G2, test(T(G1,S1)), test(T(G2,S2)),
traversal change specification
TS Comparable information
15
Example
S1
A
C
S2
A
B
C
A
B
C
K
A
B
C
G1
G2
X
K
A
B
C
A
B
C
T(G1,S1)
T(G2,S2)
AP
OO
A
B
C
Errors
A
B
C
Errors
K
K
C
C
A
B
A
B
Diff
Diff
16
Example

• S1 from A to C
• G1A,BB,C
• T(G1)A B C
• Compatible yes
• Errors no B,C
• Diff B,CgtB,K. new K,C

• S2 from A via B to C
• G2A,BA,XB,KK,CX,C
• T(G2) A B K C
• Compatible yes
• Errors no B,C
• DiffB,CgtB,K. new K,C

17
Comparison DJ versus OO during evolution

• Static analysis information
• DJ gives similar static information as OO
• Evolution effort
• DJ never worse than OO
• in many cases DJ less effort
• Danger of wrong traversals testing needed.
• OO possibility of errors not detected
  statically.
• DJ possibility of strategy being wrong. But if
  strategy correct, traversal guaranteed to be
  correct.

18
Comparison DJ versus OO during evolution

• If class graph changes, it is necessary to
  re-test the program.
• Whether program is hand-coded or specified
  through strategies, it needs to be tested.

19
Abusing DJ

• No retesting after class graph change. Reusing a
  piece of software (e.g., a strategy) in a new
  context (a new class graph) requires retesting.
  Antidecomposition adequacy rule (Weyuker)
• There exists a program P and a component Q of P
  such that T is adequate for P, T' is the set of
  vectors of values that variables can assume on
  entrance to Q for some t of T and T' is not
  adequate for Q.

P
Q
20
Application to DJ

• Adaptive program Q used in program P (determined
  by a class graph and Q).
• If P is adequately tested, Q might not be
  adequately tested.

P
Q
21
Abusing DJ

• Ignoring Errors() and Diff() output because the
  new program compiles and runs. Uncritical
  acceptance of the analogical transformation done
  by DJ.

22
Programming in DJ

• Can avoid any direct access to a part program
  becomes elastic.

23
Computing your class graph

• ClassGraph cg new ClassGraph()//.java
• // Need to do once

24
Instance-variable-free programming

• Rule for eliminating instance variable access
• // A a new A()
• // bb a.get_b()
• bb (B) cg.fetch(a, new
• Strategy(A-gtB))

25
Gathering objects

• Collecting B-objects reachable from A
• Vector bb cg.gather(a, new
• Strategy(A-gtB))

26
DJ example

• ClassGraph cg new ClassGraph()//.java
• A a new A()
• cg.traverse(a, new Strategy
• (A-gtB, new MyVisitor())