Re: A question on UML Profile and stereotype

Re: A question on UML Profile and stereotype

From: Pieter Van Gorp <>
Date: Thu, 18 Dec 2008 11:46:24 +0100
Message-ID: <>
Dear Earl,
interesting to get an update from a UML 2 user.

On Thu, Dec 18, 2008 at 11:26 AM, Waldin, Earl <> wrote:
> With respect to UML, technically speaking, a stereotype definition is
> at level M2. A stereotype is a metaclass that conceptually subclasses UML
> metaclasses. I say conceptually because the official relationship is
> extends, not subclasses, although it looks similar. From the UML 2.1.1 spec
> (section 18.3)
Strange!  As I indicated in my previous reply, the stereotype
definitions used to be at the M1 level too.  This made a lot of sense
from a tool interoperability point of view so I will check out the
most recent UML and MOF specifications to double-check your answer.

> When you apply a stereotype to an M1 class, you create an instance of that
> stereotype.
For UML 1, this used to be different too.  In UML 1, applying a
stereotype, involves creating a link between the application model
element (e.g. Car) and the element that defines the stereotype (e.g.
Entity).  There is no instantiation involved there.  Again, I will
check out the most recent UML spec...

> This is the meta-circular conceptual
> price you pay when you want something like UML to define a way to extend
> itself using its own language.
You do not pay it when using UML 1 since there is no instantiation at
stereotype application time.

I will get back to you as soon as I have drilled through the most
recent specs :-)

Pieter Van Gorp
  Assistant Professor (Universitair Docent)
  Information Systems Group
  School of Industrial Engineering
  Eindhoven University of Technology (TU/e)
  Office: Pav.D12
  P.O. Box 513, 5600 MB Eindhoven
  The Netherlands
  Phone: +31 40 247 2062
  Skype ID: pvgorp
  Fax: +31 40 243 2612
Received on Thu 18 Dec 2008 - 10:50:07 GMT