Re: UML interface element suggestion

[Kevin Page <krp ecs soton ac uk>]

On 28 Sep 2001, Lars Clausen wrote:
> > It would be useful to have a connection point added to the interface
> > object in the UML section - something so that you can draw a
> > dependency line to the circle part of the lollipop (to show that a
> > class uses that interface).
>
> That would be useful.  Looking at the overall design, I wonder why
> 'interface' doesn't exist as a separate entity, but only as 'implements'.
> Don't people use interfaces without knowing their implementation?  Isn't
> that the whole point?  So shouldn't we have an 'interface' object that's
> just a circle with name and connection points, and then an 'implements'
> connector?

<disclaimer> I don't know very much about UML. At all. I've just
started trying to use it since it seems to be a vaguely standard way
to describe software </disclaimer>

Yes, it would seem sensible that you'd need to describe an interface
without an implementation.

I think it comes down to which UML diagram type (or view, or whatever
it's called) you're using the different notation in. I haven't seen
the lollipop interface object without an implements line in any
examples I've come across.

Maybe you're meant to use the standard class entity with an
<<interface>> stereotype if there isn't an implementation to
document? (I have seen this in examples)

Regards,

kev

-- 
Kevin R. Page
krp@ecs.soton.ac.uk      http://www.ecs.soton.ac.uk/info/people/krp
Intelligence, Agents, Multimedia      University of Southampton, UK