Constraining autodefinition of Stereotypes

[Paolo F Cantoni]

If you define a new stereotype on an item, EA "obligingly" adds it to the list of stereotypes in Settings|UML...|Stereotypes.  This is useful for new users and in many cases, however this IS a downside to it.  If you type a stereotype in and misspell it, you get another (incorrect) stereotype.  If you later spot it in the list, it's difficult to find which items have had it applied to and correct it.

I'm thinking of suggesting the ability to conditionally constrain the creation of stereotypes so that the model administrator has more control over their entry onto the Stereotype list.

Something like:
New Stereotype Management:
(x) Always add to list, without asking
(  ) Ask before adding
(  ) Do not allow adding to list
Where the first entry is the default behaviour (as at present).
The second entry would present a dialog asking whether to add to the list, additionally presenting the closest matches in the existing list to allow the use to:
Definition:
(x) Add to list without additional dialog
(  ) Add to list and open Settings dialog
(  ) Replace with: [<existing list in dropdown - best match selected>]
(  ) Return me to item
Again, the first option is the current behaviour.

Before I submit the feature request, I thought I'd get input from the forum...
Thoughts?   Votes?
TIA,
Paolo

[Geert Bellekens]

Maybe the security option "Configure Stereotype" can be of help (if it indeed prohibits the creation of new stereotypes, I haven't tested that.

Geert

[Paolo F Cantoni]

Maybe the security option "Configure Stereotype" can be of help (if it indeed prohibits the creation of new stereotypes, I haven't tested that.

Geert

Most of our models don't require security (and like you, I haven't tested this option out - maybe someone who has can enlighten us).  But, in any case, what I'm trying to guard against is spurious configuration of stereotypes.

Paolo

[Geert Bellekens]

Word of the day:
Spurious:

not genuine, authentic, or true; not from the claimed, pretended, or proper source; counterfeit

 

BTW I agree with your point. It is rather annoying that sometimes there are three versions of (semantically) the same stereotype in the model. I'd love a way to avoid that.

Geert

[«Midnight»]

Please add me to that list, as well as to the supporters of Paolo's suggestion.