Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - olav

Pages: [1]
1
General Board / Some tests for OCL usage
« on: February 05, 2008, 12:52:12 am »
Hi!

There are some issues to OCL in EA which I do not understand properly. I did some tests and came to the conclusion that OCL is not properly implemented, yet. However, I would be very interested in an announcement to what extend and when new OCL features will be implemented in EA. Here is a list of tests I did:

Consider a class having an attribute minNumber of type int which has an initial value of 6. I added an OCL constraint like this to this class: "inv minNumber: self.someNumber >= 6". I hit save and EA says "OCL validation successful", which sounds cool, but seems to refer only to the correctness of the OCL syntax. If I do a model validation (only "Element: OCL Conformance" is enabled) EA says: "Validation complete - 0 error(s), 0 warning(s)". Perfect! If I change the initial value of the attribute to a value lower than 6, the model validation fails - as expected. Perfect!

However, if I change the OCL expression to "inv minNumber: self->someNumber >= 6" (please note the "->" instead of the "." after "self), EA still says "OCL validation successful", although this is not valid OCL - to my understanding. Is my understanding correct? If yes, when will the OCL syntax check be updated in EA?

Next issue: Consider the classes "LoyalityProgram" and "ProgramPartner" (not the standard example). There is an aggregation from ProgramPartner to LoyalityProgram of multiplicity 1:1, which has the role name "partners" on the ProgramPartner's side. Further, there is an attribute "value" of type int and initial value 1 in ProgramPartner.

Now I add a constraint to LoyalityProgram like this: inv valueCheck: self.partners.value = 1. The syntax is okay according to OMG's OCL specification, although "partners" is supposed to be a collection (the nit-picker's syntax should be "self.partners->collect(value) = 1"). However, for both versions EA does say "OCL validation successful", but the actual model validation fails.

I did not find any means to make EA-OCL navigate associations properly. Each time a rolename of an association joins the game, EA reports a "MVR040001 - warning...". There are two possible reasons for this: 1. My understanding of OCL is simply to "basic" so I am not able to use it properly. 2. EA-OCL is not able to navigate associations. If the latter is the case, I would love to know, when this is supposed to change.


Cheers,

Olav

2
General Board / Re: Inserting Messages in Sequence Diagrams
« on: June 05, 2005, 11:40:41 pm »
That's it!

In true sequence diagrams I am perfectly able to re-arrange the messages as I like.

In class diagrams I can drag diagram reference frames across the whole diagram (even y-dimensional) in order to make them frame that part of the diagram I want to.

There is no type of diagram that allows for both features (sequencing messages and framing diagram elements not at the top of the diagram), right?

Hence: In general, how can I link diagram elements (like messages) to diagrams the way that when I click (double-click or whatever-click) on it the referenced diagrams pops up? Is there a way to achieve this?

Thank you once more!


Cheers, Olav

3
General Board / Re: Inserting Messages in Sequence Diagrams
« on: June 01, 2005, 11:57:52 pm »
Sargasso!

Thanks for being that responsive!

I just wonder how you come up with the idea of z-ordering being of help to my efforts... Possibly my English is too cryptic after all... Or I am indeed completely off the track...

When I talk about layers of abstraction I do not mean different layers shown in a single diagram. More so I structured the overall modelling problem according to distinct layers of abstraction and want to display one single layer per diagram. Talking about sequence diagrams there are messages on a high level of abstraction that involve quite a few activities in detail. Think for example about a HTTP-Request that may be shown as a single message between the browser and server (high-level) OR numerous tiny messages between all the involved machines en route from the machine the browser is running on and the server machine (low-level).

Assuming that the word 'layer' caused you to hint at z-ordering I furtherly assume that we have a classic misunderstanding, right? I hope that the clarification above is indeed that and here I am trying to rephrase my question:

In order to link a single message of a sequence diagram to another diagram depicting the gory details of that particular high-level message, I gathered using diagram references framing the high-level message in question would be a good idea. However, in a 'real' sequence diagram (i.e. a diagram created as of type 'sequence') it is seemingly impossible to drag a diagram reference off the top of the diagram. I can only drag it to the left or right, but unfortunately not to the bottom.

Is referencing a diagram this way actually a bad idea? Is there in general a good way of linking diagrams to single elements of other diagrams?

Thank you so much for your time&effort!


Cheers,

Olav

4
General Board / Re: Inserting Messages in Sequence Diagrams
« on: June 01, 2005, 01:35:29 am »
Hi!

Thanks for the instantaneous reply! It's funny, but today I can insert messages in between without having to access the menu as you hinted me to do.

Anyway, even if I would want to 'Sequence Communication Messages' the menu item stays always ghosted in EA. What's wrong?

The sequence diagram uses 'untyped' objects, i.e. I did not specify a class for the lifelines. Is that what keeps the menu item from being enabled?

Anyway, struggling for as much help as possible I have another related question: When I insert another diagram as a 'diagram reference' into the sequence diagram it sticks always to the top. However, I would like to move the frame down a bit in order to place it around those messages that are being modelled more precisely in the referenced diagram.

When I insert a diagram reference into a class diagram it works as desired. That's what I did yesterday: I created a class diagram, drew a sequence diagram in there and tried to insert messages somewhere in between. I know this sounds crazy (it is crazy actually, I gather), but I made progress - at least until I wanted to insert messages somewhere in between.

So my 'formal' requirement is: I want to draw a top-level sequence diagram that references more detailed sequence diagrams. I want to be able to navigate between the different layers of abstraction by means of (double-)clicking on certain diagram elements.

Is my approach completly off-roads? Could you or somebody else give me a hint or two, please?

Thank you very much!


Cheers,

Olav

5
General Board / Inserting Messages in Sequence Diagrams
« on: May 31, 2005, 05:18:46 am »
Hi!

I just tried to update a sequence diagram of an existing model. The issue is to insert some messages somewhere in between. Unfortunately, I failed miserably...

It seems to me that it's only possibly to append new messages to the end of a sequence diagram, but I hardly can believe this to be true. Could anybody give a hint about how to do that if that's possible at all?


Cheers,

Olav

6
Hi!

Parsing the EA model via COM yields a problem. When I want to have a look at the layout of the diagrams I get the collection of displayed objects (DiagramObjects) and have a look at each individual DiagramObject. These have the members left, right, top, and bottom which (according to the docs) ought to hold the position of the object in the diagram. Unfortunately, the values found there do not correspond to the actual position I perceive by looking at the diagram.

What's wrong with my approach? What am I missing?

Thank you very much!


Best Regards,

Olav

Pages: [1]