Will the UML ever become truly universal?

A broad description of UML may be found at
http://en.wikipedia.org/wiki/Unified_Modeling_Language
The following passage is relevant to this message:

Models vs. diagrams
It is important to distinguish between a UML model, and
a UML diagram, or set of diagrams, including Use Case
Diagram, Collaboration Diagram, Activity Diagram,
Sequence Diagram, Deployment Diagram, Component Diagram,
Class Diagram, StateChart Diagram--a UML diagram is a
graphical representation of the information in the
model, but the model exists independently. XMI in its
current version provides interchange for any OMG models.
Diagrams can also be represented as models using the
Diagram Interchange (DI, XMI[DI]) standard, but since
it is relatively new very few tools support it.

A broad description of XMI may be found at
http://en.wikipedia.org/wiki/XMI
but to cut to the chase it is the third paragraph which is
the most striking in respect to this message:

Several versions of XMI have been created: 1.0, 1.1,
1.2 and 2.0. The 2.0 version is radically different
from the 1.x series. However it brings XMI closer to
idiomatic XML usage, but is unfortunately very rarely
supported by modeling tools. The lack of good XMI
support is slightly incomprehensible, since it is not a
particularly complicated standard to adhere to. Then
again good interoperability might not be in the best
interest of the tool vendors themselves due to "vendor
lock-in".

Which brings me back to the (provocative) subject line.
Surely it is lack of interoperability that prevents UML
from being truly universal? From a user's perspective the
answer to that question must be yes. If I spend (a lot of)
time in an independant UML tool or a programming language
which has an integrated UML tool, it would surely not be
unreasonable of me to expect to export and import my work
(model) into another tool or comparable programming language
--regardless of vendor or programming language (C++ to Ada,
for example). Besides, a product may be unavailable on
certain system platforms. Clearly this is more than a
matter of freedom of choice, however important that may be.

IMHO, interoperability should be a requirement of the UML
specification, for without it the universality of UML is
severly hampered.

--
Mail sent to this email address is deleted unread
on the server. Please send replies to the newsgroup.


.

Relevant Pages

  • Re: Coding monsters and items
    ... What UML models are object and their realtions. ... probably just use one diagram and mix in elements from other diagram ... a series of steps that occur in the game. ... It's a distinct game design element. ...
    (rec.games.roguelike.development)
  • Re: round-trip engineering
    ... the diagramming syntax is not UML. ... But I would like Visio to have round-trip engineering which is ... And if user changes UML diagram info, ...
    (microsoft.public.visio.developer)
  • Re: Place objects in threads and processes in UML
    ... Since threads are highly dependent on the specific implementation, one generally does not deal with threads in UML models. ... However, when the application space /is/ the implementation space subject matter, as in your framework, one is in a different ball game. ... That diagram just shows the object interactions that are relevant to the thread. ... Another approach is to use the Package Diagram to collect model elements in thread packages. ...
    (comp.object)
  • Re: un-group and un-constrain all the class diagram elements
    ... Visio diagram. ... The destination diagram is not UML, ... connection end labels around and I cannot change block height. ... You might find that Visio's built-in UML is causing some of the ...
    (microsoft.public.visio)
  • UML and fp
    ... I spent a week on an UML course because my colleagues expect me ... to diagram the stuff I code and UML is the company standard. ... He said, you don't, there is no UML data flow diagram. ...
    (comp.lang.functional)