Subject: C441 JGSI Review Resent-Date: Thu, 30 Sep 1999 22:04:50 -0400 Resent-From: Geoffrey Fox Resent-To: p_gcf@npac.syr.edu Date: Fri, 10 Sep 99 9:22:20 MET DST From: Henri Bal To: gcf@npac.syr.edu Reviewer: Thilo Kielmann, Vrije Universiteit Amsterdam ========================================================================= Review of: Design of the Kan Distributed Object System a)Overall Recommendation reject b)Comments suitable for authors I recommend to reject this paper for two reasons: 1. I could not find anything new in here except that all those well-known concepts have now been implemented in/for/with Java. 2. The paper lacks proper organization, structure, and most of all focus. Furthermore, I strongly disagree with the authors that asynchronous RMI is equivalent to creating a new thread and later synchronizing with it. In fact, that's what you can already achieve with Java's synchronous RMI. What would really be interesting (but still nothing new) would be an async. RMI that avoids multithreading overhead at all. I consider this to be a flaw of the paper. Also, the paper is quite hard to read. As an example, take the first paragraph of section 2.1. I finally gave up trying to understand it. I strongly advise you to extract your contributions and to rewrite your paper around them. Please try to remove everything that deviates from your paper's focus (e.g. the discussion of implementation details in the introduction). Cleanliness of programming models is hard to sell. In this paper, you failed to do so. Please come up with convincing examples of how your model achieves your goal of "clean expressions of concurrent algorithms". In Fig.2 you give an example of how async RMI should not been used. Why don't you give a positive motivation? Esp. Section 4 needs a clear structure and the removal of unnecessary detail like the table that compares various ways to read the clock. In your "related work" section, you need to emphasize why Kan is better than you competitors instead of the neutral comparison of "we do it this way and they do it that way". That closes the loop over your paper: What is new in it? c)Comments to the editor