Demeter Seminar (6. meeting): Place: Cullinane 107 Date and Time: Nov. 30 at 2pm: Topics: Linda Seiter and Karl Lieberherr will present: Progress on Adaptive Behavioral Components and Report on the OOPSLA 1995 workshop on Adaptable and Adaptive Software The workshop report will appear in the addendum to the OOPSLA '95 proceedings. ============================================== Demeter Seminar (5. meeting): Place: Cullinane 107 Date and Time: Nov. 9 at 2pm: We continue our discussion of: How should the next generation of adaptive software look like? Topics: Solution to the adpative iteration problem Integration with the Unified Method of Booch/Rumbaugh/Jacobson Progress reports on papers under preparation (Linda and Salil) - ABCs - Codifying Design Patterns using Adaptive Programming At 3pm, Remy Evard will join us for a discussion of tools for scientific collaboration which would help our research in Adaptive Software. Karl Lieberherr PS: Next week is no seminar because Mitchell Wand, Boaz Patt-Shamir and I need to finish an ARPA proposal on Adaptive Programming. ================================================== Demeter Seminar (4. meeting): Place: Cullinane 107 Date and Time: Nov. 2 at 2pm: How should the next generation of adaptive software look like? Recent research in the Adaptive Programming Laboratory by (alphabetically) Walter Huersch Crista Lopes Jens Palsberg Boaz Patt-Shamir Salil Pradhan Linda Seiter and myself leads to a new vision of how an adaptive programming system should look like. Let's put the pieces of the puzzle together. There are many open research questions. ================================================= Demeter Seminar (3. meeting): Place: Cullinane 107 Date and Time: Oct. 26 at 2pm: Neeraj Sangal, Visiting Professional, Northeastern University -------------------------------------------------------------------- This seminar will describe the work done by Don Batory et. al. at the University of Texas at Austin on the GenVoca style of software design for building domain specific software system generators. We will focus on their use of the GenVoca system for building a data structure generator. We will also discuss the motivation for this undertaking and whether there are any comparisons to be made to Adaptive Software. ==================================== Demeter Seminar (2. meeting): Place: Cullinane Hall Room 8 (not 107 this time) Date and Time: Oct. 5 at 2pm: Oct. 5: Cristina Lopes (Northestern University and Xerox PARC) A New Semantics for Adaptive Programs Adaptive software is a generic idea which can be realized in many different ways. One possible realization is in the Demeter Tools/C++ as implemented by Cun Xiao and partially formalized in the TOPLAS paper by Palsberg, Xiao and Lieberherr (see chapter 15 in the Demeter book). A second realization by Walter Huersch and Linda Seiter is in ftp://ftp.ccs.neu.edu/pub/people/huersch/papers/HK95-oo-evolution.ps.Z A third one will be presented by Cristina based on her work at Xerox PARC over the summer. Cristina started with Walter's and Linda's approach and then developed a new semantics. I expect we will have a lively discussion as usual. -- Karl Lieberherr PS. Good news: Two new papers have been accepted at ISOTAS '96 in Japan: The conference version of: ftp://ftp.ccs.neu.edu/pub/people/huersch/papers/HK95-oo-evolution.ps.Z and ftp://www.ccs.neu.edu/pub/people/crista/isotas96.ps Congratulations to Crista, Linda and Walter. ========================================== For information about the seminar and on previous and future talks, see: http://www.ccs.neu.edu/research/demeter/seminar/seminar.txt (future talks are mentioned at the end.) =========================================== 1. seminar Welcome to the Demeter seminar: We will meet this quarter on Thursday afternoons from 2-4 in 107 Cullinane Hall. We start on Sep. 28. ------ The plan for the seminar is that advanced Ph.D. students and faculty or industrial experts introduce new Ph.D. and advanced master students and guests to research topics in adaptive software and related fields. The prerequisite for the seminar is to take COM 3360 concurrently or to have it taken earlier. (The textbook for COM 3360 Title: Adaptive Object-Oriented Software: The Demeter Method with Propagation Patterns Author: Karl Lieberherr Publisher: PWS Publishing Company ISBN: 0-534-94602-X Year: 1996 is available at the NU Bookstore.) Why adaptive software? It is an important topic in computer science and one which was developed at NORTHEASTERN. We are in a unique position to contribute to this area since we know it so well. Why is it important? 1. One indicator is in /ccs/ftp/pub/people/lieber/Demeter-interest 2. Another one are the position papers accessible through URL: http://www.ccs.neu.edu/research/demeter/adaptable-systems 3. Another one is that about $836 000 grant money from the National Science Foundation and from companies such as IBM and Mettler-Toledo have been invested into this area. ------ The first presentation is by Linda Seiter-Keszenheimer on: Managing Adaptive Components during Evolution Date and Time: Sep. 28, 2-4 Location: Cullinane 107 Object-oriented software engineering is becoming increasingly popular for system development, primarily due to the production of reusable components such as class libraries. Each time a component is reused, the issue arises as to how to verify or test the component in its new environment. A well tested component may require additional testing when it is reused, since errors may be introduced when it is placed in a new context. The foundation of an object-oriented system is the class structure, upon which program functionality is built. The class structure also constrains the objects stored in a database. Changes to the class structure can require significant maintenance of the persistent object store and existing programs, as well as significant retesting. {\it Adaptive software\/} describes a software model where functional components are developed with the intent of being flexible to changes in the class structure, allowing an adaptive program component to be reused with many class structures. However, while an adaptive component may work correctly with one class structure, it could behave erroneously with another. This presentation outlines a framework for managing adaptive object-oriented software during evolution. Software evolution is separated into two distinct processes: structurally motivated transformations and behaviorally motivated transformations. A structural transformation modifies the class structure. Existing programs and objects may become inconsistent with the evolved class structure. The inconsistencies are identified and the necessary object and program transformations are applied to reinstate consistency, while maintaining the original behavior of the system. Behavioral transformations may subsequently be introduced based on the modified class structure. The framework will define the maintenance, verification and testing process through each stage of development and evolution. The goal of the framework is to minimize the effort required to maintain and test a system as it evolves. {\bf Keywords: } Object-Oriented Software Engineering, Software Testing, Software Maintenance, Schema Evolution, Verification and Validation. ===================================== Planned presentations: Oct. 12: Neeraj Sangal (Northeastern University) Review of some of the work of Don Batory's group at the University of Texas at Austin Oct. 19: OOPSLA '95, no meeting. (The Demeter Research Group organizes a workshop, has a demonstration of the Demeter Tools/C++, Linda Seiter's research was selected for the Ph.D. forum and Crista Lopes has a poster on Adaptive Parameter Passing.)