Automated Software Design
(formerly Feature Oriented Software Design FOSD)
CS392F and EE382V
Fall 2013

	Instructor	Prof. Don Batory	batory@cs.utexas.edu
	Office Hours	GDC 5.826  Tuesday 3:30-4:30pm
	Room	GDC 2.210
	Time	MW 3:30-5:00pm
	Unique#	CS 54123 (ECE 17294)
	Ground Rules	Rules of Conduct and Academic Honesty
	Your Grades	UT Blackboard Grade Book
	Final Exam	TBA.

• Generative Programming — programs are synthesized by other programs
• Compositional Programming — programs are assembled by composing prefabricated parts
• Domain-Specific Languages (DSL) — domain-specific notations that simplify program specification
• Model Driven Engineering  (MDE) -- programs are specified as and synthesized from models
• Automatic Programming — efficient programs are synthesized from declarative specifications
• Program Analysis — tools that evaluate properties of programs

A spectacular example of the integration of these areas was realized thirty-five years ago: relational query optimization (RQO). A relational query is specified in SQL, a parser maps it to an inefficient relational algebra expression, a query optimizer optimizes the expression automatically, and an efficient query evaluation program is generated from the optimized expression. SQL is a prototypical declarative DSL. Query evaluation programs are specified as compositions of relational algebra operations; relational algebra is a prototype for compositional programming. Query optimizers achieve automatic programming by rewriting an inefficient expression/program to a semantically equivalent but more efficient expression/program. The cost models that drive expression optimization are examples of program analysis. Mapping a relational algebra expression to an efficient program is generative programming and is an elementary example of model driven engineering.

A "holy grail" of Software Engineering is to replicate the success of RQO in other domains. Feature Oriented Software Development is a generalization, and its ideas are at the confluence next-generation research topics in software modularity, program design and program synthesis: OO design, product-lines, program refactoring, model driven engineering, program evolution, and program transformations.

Prior offerings of this course lead to student publications and research degrees (M.Sc. and Ph.D), some -- not all -- are listed below:

• S. Trujillo, D. Batory, and O. Diaz. Feature Oriented Model Driven Development: A Case Study for Portlets. International Conference on Software Engineering (ICSE), 2007.

• S. Apel. How AspectJ is Used: An Analysis of Eleven AspectJ Programs. Journal of Object Technology, 2008.

• G. Freeman, D. Batory, G. Lavender, and J.N. Sarvela. Lifting Transformational Models of Product Lines: A Case Study. Software and Systems Modeling. October 2009. Best papers in ICMT'08.

• B. Delaware, W. Cook, and D. Batory. Fitting the Pieces Together: A Machine-Checked Model of Safe Composition. ACM SIGSOFT FSE, August 2009.

• B. Delaware, W. Cook, and D. Batory. Product Lines of Theorems.  SPLASH/OOPSLA Conference, October 2011

• B. Marker, A. Terrel, J. Poulson, R. van de Geijn, and D. Batory.  Mechanizing the Expert Dense Linear Algebra Developer (Poster Paper), PPoPP, February 2012

Prerequisites, Overview, and Goals

All programming assignments will use Java or Prolog. I expect proficiency in Java; you can pick up what you need in Prolog. We will use the following software, all of which is free to UTCS students, and all of which has been installed in the UTCS Microlab, which is in ENS 1 & 2, and PAI 5.38.  The following is for Windows Platforms.  If you use Apple machines, well, I will try to help as much as I can, but no promises!

• NetBeans IDE 7+  -- I prefer NetBeans to Eclipse for program development as it is a simpler environment. 
  

• Microsoft Visio 2010/2013 -- for drawing UML diagrams.  If you are a CS student, you can gain access to Visio via Software Downloads, read the MSDN/AA agreement, click the secure download site link, and follow the instructions. For ECE students, do the same at the ECE-MSDN/AA web page.

• VLC Media Player -- to view .MOV videos shown in class.  (You can use Windows Media Player, but its video resolution is awful).  No need for this software if you are a Mac user.

• SWI Prolog -- for writing metamodel constraints
  

Basic familiarity with the following topics are assumed -- all that is needed will be covered, but some prior experience will help greatly:

• Java
• Compilers and Grammars
• Object-Orientation
• Relational databases

As mentioned above, we will use Prolog in assignments. I expect that you'll pick up what you need.

Lecture notes will be presented online (after the lecture) as downloadable PPTX files.  Links to the lectures are given below in the Course Outline. There is an optional text for this course:

• B. Pierce, Category Theory for Computer Scientists

• 10% class participation, quizzes, homework
• 20% programming assignments
• 30% project and presentation
• 40% Midterm and Final, with the final weighted a bit more

Each group of students will complete an approved project by the end of the class.  A classroom presentation on every project is expected.  Details on the projects will be announced later.

• Submitting homework or program HP: create a zip file <yourname>.zip, and run on a linux box:
  
  > turnin -- submit dsb HP <yourname>.zip
  
  where HP is a designator H1, ... for homeworks and P1... for programs
  
• A PDF file must be included on all assignments -- it lists your name and email addresses that are hyperlinked (so that I can easily send my comments of your assignment back to you).

Numbers in [brackets] indicates the estimated number of lectures on a topic. The number indicated is a lower-bound, as there will be class room discussions to work on problems and review of homework assignments. Papers in yellow are required readings and are accessible via a web link. Required papers are to be read prior to each class. The order in which topics are presented might be changed as the class progresses.  (#) indicates approximate number of class dates per subject.  indicates code examples can be downloaded. 

Be aware: I reserve the option to replace a series of lectures below with another that I feel may be better and will improve the course further.