"NOT allowed" relationships in QuickLinker

[Paolo F Cantoni]

The QuickLinker is a NEAT feature with a lot of potential!  It can really ease the burden of model creation and greatly improve consistency.  Sometimes, particularly between specific types of elements - you'd like to be able to say:  "Hey, there should be NO relationships (at least of this type) between these two types of things...

It seems to me that by adding something like:  New Link Direction = none
you could specify that the QuickLinker should display the value of the New Link Caption column and suppress all (other) linkages of that type between these two metatypes.

Further, if the New Link Type could be set to <all>, you could effectively suppress all types of relationships between the two metatypes.

Feature Request submitted (under this title)

If you want to vote for this, submit a Feature Request with the same name,
leaving a note here to say you've done it
- so we users have an indication of the level of support.[/color]

Paolo

[SomersetGraham]

Feature Request sent

[Oliver F.]

Isn´t what you want to accomplish done via a simple line in the CSV with only A-D being filled and "Exclusive to ST Filter + No inherit from Metatype" set to true?

Oliver

[Paolo F Cantoni]

Isn´t what you want to accomplish done via a simple line in the CSV with only A-D being filled and "Exclusive to ST Filter + No inherit from Metatype" set to true?

Oliver

Hi Oliver,

In toto, NO... since I'm eventually asking for this to suppress across QuickLinker specifications.

But even in partem, I believe "Exclusive to ST Filter + No inherit from Metatype" set to true only inhibits the generic (Help: do not display the Quick Linker definitions of the equivalent unstereotyped element.) But I may be wrong in interpreting the Help.  I'll try it out and get back here.

In any event, I want (if possible) to actually inform the user that the linkage is illegal; not that I've done something wrong with my QuickLinker definition - which will ALSO suppress QuickLinker entries...    (and which WON'T inform the user...)

Paolo

[«Midnight»]

Yes Paolo!

Regardless of the results when you check this out...

Your second point hearkens back to basics. If you do not document your work (such that the end user can understand what is actually happening) then you have not done your work correctly. If you cannot document your work then you cannot do your work correctly. If the Quick Linker prevents you from (simultaneously) doing the job right (preventing an improper link) and telling the user what is happening (that the link is improper) then something is just plain wrong.

David

[Oliver F.]

Isn´t what you want to accomplish done via a simple line in the CSV with only A-D being filled and "Exclusive to ST Filter + No inherit from Metatype" set to true?

Oliver

But even in partem, I believe "Exclusive to ST Filter + No inherit from Metatype" set to true only inhibits the generic (Help: do not display the Quick Linker definitions of the equivalent unstereotyped element.) But I may be wrong in interpreting the Help.  I'll try it out and get back here.

From what I recall you can create a linker definition from element a stereotype A to element b stereotype B and leave everything else empty except the "Excelusive to ST Filter " column which will prevent connection options shown from a to b (with the corresponding stereotypes).

Oliver

[«Midnight»]

AFAIK that is correct Oliver. It's been some time since I set up a QL profile but that seems familiar.

I think one of Paolo's points - the one I was commenting on - was that this is fine as far as it goes, but not really enough. Instead of (just) disallowing the link the QL should give the user some kind of feedback that such a link is not allowed.

At the bare minimum that a <link type> is inappropriate or invalid here. And <link type> should allow or include <link stereotype> if defined in a profile. It would be better if the QL profile provided a default message and a way that the profile designer could override this with a custom message for this circumstance. Better still would be the ability to specify that <link type> or <link stereotype> or perhaps some other attribute would appear at a certain point by attribute name rather than hard-wiring a value into the message.

Do I have this right (as far as I am taking it) Paolo?

David

[Paolo F Cantoni]

David,

"I like the cut of your jib, Sir!"

Yes, you're on the same track...

To both of you, I checked my QL and ALL the settings have "No inherit" set to TRUE so it doesn't do what I need...  Yes, It suppresses the generic linkages, but doesn't inhibit the creation of other linkages.  Indeed, since I'm just (in the current implementation) defining another relationship., if the user selects the message, they actually get a relation.

In addition, it can't suppress across QL definitions.

That's why this is a Feature Request.

Paolo