Hi Binoy: >From binoy@ccs.neu.edu Thu Sep 18 14:54:56 1997 >Date: Thu, 18 Sep 1997 14:51:49 -0400 >From: Binoy Samuel >CC: dem@ccs.neu.edu >Subject: Re: UML and cds in AP Studio > >Karl Lieberherr wrote: >> >> Hi Binoy: >> >> What is your current view of: >> >> $DEMETER_HOME/demeter-method/uml-syntax-discussion >I for one prefer the syntax in the .cd. It feels like a >natural extension, and is reflected in the ApStudio drawing process too. > >> What happens to a cardinality 3..4? > >Currently such a cardinality would only remain upto the design phase. >ApStudio wouldnt support that(right now). However as and when we extend >class dictionary support, it wouldn't be a big deal to support it in the >translation process(cd-uml-cd). >What kind of code generation would this lead to? I wonder if it would be >worth the while to implement such constraints. I agree with you that this probably would not be worthwhile. But some information systems people might disagree. We could eventually generate a demon which checks the constraint (class invariant). > >> >An inheritance from a non-abstract class would be achieved using an >> >"extends" edge in ApStudio. This would translate to something like >> > A = .. *extends* B. (B is the non-abstract class). >> >> Are the traversals working through such edges? >No. And since the subgraph computation will take place through the Ok. That is good to know. To make it clear: Inheritance edges expressed with *extends* are ignored by traversals while inheritance edges expressed with A : B | C ... are taken into account. >common library implemented by Doug, there would be uniform handling >allthrough. > >> >> It is important that we support standards like UML in there generality >> in the long run. Otherwise people think that we have only a toy OO model. >I think we should maybe have a detailed session on what can be done with >UML, and what we are currently supporting. > >-binoy > Let's do this at our next seminar. -- Karl