Re: Fortran bashing in the MATLAB newsgroup

On 30 jun, 14:52, Damian <dam...@xxxxxxxxxx> wrote:

Interesting points about conflicts between OCL and UML. I'd love to
hear more.  On a similar note, I've sometimes had difficulty
expressing Fortran concepts in UML.  For example, UML has no concept
of multi-dimensional arrays.


Well, here are a few, perhaps minor, details:
- Where UML talks of Classifiers, OCL talks of types:
class, interface, datatype or component.
- UML does not specify how constraints on a superclass are
assed on to the subclasses. OCL needs to know that, so some
practical rules are suggested.
- OCL has no real way to deal with inheritance, whereas this
seems very important in UML.
- OCL can treat any attrribute as a collection, whether it is
one in the UML diagrams or not.

(The book I refer to is that by Jos Warmer and Anneke Kleppe)

These are just my impressions, but as they claim that with
UML and OCL you should be able to generate the "complete"
source code for the system, I wondered about the ambiguities
and apparent mismatches I came across.

Regards,

Arjen
.

Relevant Pages

  • Official status of UML2?
    ... OMG says latest UML version is UML 1.5: ... yet many some articles I come across imply UML 2 "is out". ... OCL 2.0 is shown as "adopted". ... Diagram Interchange are in "finalization started" status. ...
    (comp.object)
  • Re: Fortran bashing in the MATLAB newsgroup
    ... Interesting points about conflicts between OCL and UML. ... expressing Fortran concepts in UML. ... The typical data structures I find in Fortran programs or general ... In a cynical mood I might say: one-dimensional structures for ...
    (comp.lang.fortran)