Standardising Classifier adornments

[Paolo F Cantoni]

Bruce (Sargasso), in a private message, concerning another post of mine,
made the following comment (which he has permitted me to quote):

I spend, I reckon, at least a day on each project telling
modellers not to use their own ideas of what's beautiful and
meaningful - and then hacking into the EAP repository to cut out
their beautiful pictures when they ignore me and do it
anyway.

Bruce, here, is talking about the ability to replace the entire
model element (typically a shape only - why?) with an image.
Thus for the actor stereotype we replace the box with a "stick
man".

One of the functions of the model is to communicate, and from
time to time in the past I've used this technique to better
communicate some concept or other to the "punters" - the
stakeholders, users, management, development teams etc.

However, if you observe the UML specifications carefully, you'll
see that "all boxes are equal - but some are more equal than
others".  Some boxes are lucky enough to have icons in their top
right hand corner - Component for example.  I can't quickly see
other examples - and I can't use EA as it isn't always UML
compliant here.

It seems to me that all classifiers should have the ability to
be adorned with icons in the top right hand corner.  Thus, for
example, a boundary entity should have a number of rendering
modes:  Plain "box", Plain box with icon in top right, Icon
replaces box.

We can then extend this to the application of stereotypes, we
can then get:

Standard or icon adorned box with «stereotype» added as text,
Standard or icon adorned box with «stereotype» images added as
icons in top right (or perhaps top left), principal «stereotype»
image replacing box with subordinate «stereotype» text
adornment, principal «stereotype» image replacing box with
subordinate «stereotype» images added as icons (as above).

In this way, we can (possibly) better communicate the
application of multiple stereotypes to a model element.  I
believe multiple stereotypes will become increasingly more
useful as people discover they are possible (compared to single
stereotype UML 1.x)

Thoughts anyone?

Paolo

[sargasso]

The concept of fully replaceable "boxes" would allow us to cope with whichever version of UNL is desired for the particular environment.

For example,  if a client is refraining from moving to UML2 then the UML 2 component box as introduced to suit a certain floral modelling tool's inability to resize the old component box correctly, then WE could still comply with that organisation's requirement by selecting UML 1 component boxes for their projects.

An interesting and useful proposition.  Good on ya Paolo!

I would like a caveat though - I, as the boss honcho around here want the ability to constrain the monkeys down the tree from using "pretty" but non-compliant replacement icons, replacement boxes and TRHC icons unless I SAY IT'S OK both at the project and diagram level!


bruce

p.s. Paolo - have yuou seen a spell checker for Firefox?

[Paolo F Cantoni]

p.s. Paolo - have you seen a spell checker for Firefox?

I believe GNU ASpell can be made to work, but I don't know that for sure...  Look on the site.  If you find one, let us all know as I'm seriously thinking of Firefox.

(I'll reply to the rest in a separate post)

Paolo

[mikewhit]

Even spellcheckers don't protect against homophones or near-homophones (e.g. lose/loose, defuse/diffuse, there/their/they're, fazed/phased ...).

I guess a good checker should also be aware of these !

[thomaskilian]

p.s. Paolo - have yuou seen a spell checker for Firefox?

https://do-not-add.mozilla.org/extensions/moreinfo.php?id=485is not quite what you're after, but the best I could find...

[thomaskilian]

...
I would like a caveat though - I, as the boss honcho around here want the ability to constrain the monkeys down the tree from using "pretty" but non-compliant replacement icons, replacement boxes and TRHC icons unless I SAY IT'S OK both at the project and diagram level!

I'd like to extend that request to enable the boss for limiting all sorts of extensions, including especially TAGs.

[mikewhit]

Would be 'convenient' if these customisation points could appear on a 'virtual diagram' which could then be Locked !

[Paolo F Cantoni]

I would like a caveat though - I, as the boss honcho
around here want the ability to constrain the monkeys down the
tree from using "pretty" but non-compliant replacement icons,
replacement boxes and TRHC icons unless I SAY IT'S OK both at
the project and diagram level!

Bruce

I think this is a straight forward security issue.  You should
probably start a separate thread.  Although I suspect there
could be a general pattern that could be pretty universally
applied to EA elements.

You point about "down the tree" is well taken.  It seems to me
there ought to be the equivalent of Cascading Style Sheets for
element adornments. As you specialise stereotypes you could
maintain some degree of adornment consistency - something like
the way MS Word manages style hierarchies.

Paolo

[mikewhit]

something like the way MS Word manages style hierarchies

But working.