Re: UML-Conformity

[Andrew Ross <aross westnet com au>]

On Sat, 2004-06-19 at 02:51, RittervomNie web de wrote:
> > Why would you need the text element, when the only issue with
> > multiplicities, roles, message names etc is their poor automatic
> > placement? The two bugs I previously mentioned (65430 and 118313) raise
> > two different approaches to solving this problem (improve automatic
> > placement, and provide a manual placement option), and it looks both
> > will (eventually) be implemented - yay!
> 
> Hooray for bug sollutions. I'm quite correct when it comes to multiplicities. 
> Only a hollow or filled diamond is not enough for me, I want zeros, n-s and 
> m-s and that stuff.

I'm not entirely sure what you mean here. Dia's multiplicity support in
UML Associations is string-based, so you can freely enter "*", "0..1",
"1..5", "yo-mamma" etc

>  Imagine some for-loop in sequence diagrams, messages with 
> repetition counts, if-s and stuff. When using actual versions I have to 
> emulate this by text elements to give a complete description of sequences or 
> entity-container-relationships...

I can see why loops in sequence diagrams currently need to be created
using the text tool and a rectangle (I wonder how hard it would be to
create a new UML object for this).

For messages with guards and/or repetition counts, can't you simply
include it as part of the message name? Or were you after a separate
field?

> Poor alignment makes a diagram unreadable. Multiplicities drawn within classes 
> are not quite good. And messages with off-cut underlength letters 
> (underscore, characters like g) are also hard to read. So I need a 
> workaround. Using the text element is the only way...until these bugs will be 
> fixed.

Yeah, I've been using the same work-around (text tool) here :-)
Currently I'm using the text tool for multiplicities, role and
association names, messages - I don't use any of the built-in stuff at
present (because of the alignment problems).

Cheers

Andrew