Re: [COMPONENTS] Gnome's OAF allows `Capability-based activation'...

On Mon, 10 Sep 2001, Henrik I. Christensen wrote:

> One more throught on components, ... It seems to me that we are
> talking about different things in the components/module discussion. In
> the "traditional" computer science literature components are units of
> code/functionality such as computation of a Jacobian. This way a
> components is like a java bean in the sense that it is a well defined
> box with given characteristics. At the same time we also have
> "Modules" i.e. units of functionality for a robotics system, which is
> a well defined unit in a distributed system, where we expect to have a
> particular functionality as discussed by Sara at the kickoff
> workshop. For the Modules we have particular patterns in mind such as
> the FSA model. This might not always be true for components. Thus we
> might need to have a separation between components (as specificed by
> Anders) and Modules as presented by Sara. In my mind we need a module
> definition as specified by Herman (a few night ago), but also
> components spec for chuck of code with a well defined functionality,
> that may be used for composition of modules. Makes sense?

Makes sense! We just have to decide on what word we will use for what
concept. I'll try to adapt/clarify the definitions on the web page,
but concrete suggestions are always welcome.

About modules: I've found papers that use "module" in another meaning,
i.e., as a ``compound class'' in an object-oriented design. That
means: a (possibly complex) data abstraction, but not really something
that encapsulates components (because the concept of a process or
thread within the thing is not there at all). I must confess I am
confused about the term "module"...

Herman