Hi Luis: "it contains no "pollution" from other application domains/concerns" summarizes it well. Johan was pointing out some important differences between APPCs/CGVs and frameworks: Frameworks APPCs/CGVs ----------------------------------------- A few big ones many small ones hard to combine easy to combine hard to map easy to map -- Karl PS. The GOF book contains a pretty good definition of frameworks. >From lblando@gte.com Tue Nov 10 11:00:44 1998 >To: Karl Lieberherr , lblando@ccs.neu.edu >From: Luis Blando >Subject: Re: framework reference >Cc: johan@ccs.neu.edu, lorenz@ccs.neu.edu, mira@informatik.uni-siegen.de > >At 10:49 AM 11/10/98 -0500, Karl Lieberherr wrote: >>Hi Luis: >>in your thesis you say that a framework is written for an >>ideal class graph similar to an interface class graph. >> >>Is there a book or papers that explain this approach? > >It is probably a bad use of words. By "ideal class graph" I simply mean >that the framework developer designs a framework and places the hotspots >(usually abstract classes) wherever he wants. In this sense, it is "ideal" >since it contains no "pollution" from other application domains/concerns. >In short, the point of writing a framework in the first place is exactly >the ability to write it somewhat "abstractly" so you have more freedom to >get the design uncompromised. The hotspots are used to link it to an >application later (where you "de-idealize" the frameworks class graph). > >That's all. Thus, there's no special "approach" in my statement above. It >is simply, IMHO, due to the nature of frameworks that the class graph they >end up with is somewhat "ideal". > >Let me know if you need more information. > >Luis >------ >Luis Blando, PMTS 781.466.3296 (v) lblando@gte.com >OSL, GTE Laboratories 781.466.2941 (f) 1542031@gte.pagegate.net >Waltham, MA 02454 800.971.4118 (p) www.ccs.neu.edu/home/lblando >