Position Statement Education in Automated Software Engineering

1
Position Statement Education in Automated
Software Engineering

• Professor John Grundy
• Dept. Electrical and Computer Engineering and
  Dept. Computer Science
• University of Auckland, New Zealand

2
What is the approach of our CompSci/SE Education?

• Engineering is about building models of
  products/processes (in SE some happen to be
  executable like code)
• Disciplined application of formal, semi-formal,
  and even informal techniques/technologies are
  needed in order to develop the right models
• Mix ( balance) of theory, practice, experience
• Obtain deep understanding coupled with practical
  skills in applying CompSci/SoftEng approaches
• Critical (self-)reflection continual enhancement
  of knowledge and skills quality/cost/timeliness
  focus
• Educating tomorrows practitioner today whats
  the world of SE going to be like in 2 5 10
  years time??

3
Trends

• Recent SE trends likely to continue for some
  time
• Model-driven development
• Integrated IDEs (VisualStudio, Eclipse )
• Agile (lightweight) methods
• Service-oriented architectures
• Aspect-oriented and other cross-cutting
  techniques
• Adaptive, autonomous, agent-based, intelligent
  systems
• Reliance on teamwork/shared artefacts
  supporting methods and tools
• Ever-increasing quality demands usability,
  reliability, scalability, security, trustability,
  
• Ever-increasing user base demands on software
  ubiquitous computing, safety-critical systems
• We have to educate/train students to be able to
  build/do these things

4
What is Automated Software Engineering?

• Generative try and generate code/configurations/
  other models from high-level, abstract
  descriptions (models)
• models any representation of software at
  some level of abstraction - logics, UML, MOF,
  BPML, XSLT, UIs, DBs, code,
• Component/part/service-based build
  applications from bits approach
• Ultimately autonomous agents composition and
  emergent systems (not just software!)
• Adaptive components/agents might discover
  environment and adapt to the circumstances they
  find themselves in
• Dynamic ideally can do the above at run-time
  while the software is being used
• Formalisms are necessary specifications we can
  reason with generate code from verify vs
  validate models/code

5
Do we need Automated Software Engineering?

• Of course we do! ?
• (We need an excuse to hold more ASE conferences
  after all)
• In many SE application domains we now couldnt
  get by without it
• This trend is almost certain to continue, so
• Instead of How should we educate young Computer
  Scientists/Software Engineers in the field of
  Automated Software Engineering?, perhaps we
  should be asking ourselves How much should/need
  we educate them in NON-automated SE
  techniques???

6
ASE Education

• Undergraduate SE/CS students need exposure to
• Developing models of systems software
• Verifying model properties fixing wrong models
• Generating systems from models statically and
  dynamically
• Extending generated parts statically and
  dynamically
• Composing parts (in sensible ways) statically
  dynamically
• Adaptive, dynamic, service-oriented architectures
• Automated testing techniques proactive
  reactive
• Practical, realistic applications using these
  techniques
• (Mike, can I have the Mars buggy code please?)
• BUT
• How do they learn the right time to use these?
• How do they know if ASE tools really work?!
• And how do they fix them/extend tools when they
  dont?
• Do they understand that there is no silver
  bullet? ?