Messages

376

General Board / Re: Is Sparx Asleep at the wheel ?

« on: April 02, 2019, 08:23:37 pm »

I suspect massive code refactoring would be needed, a big expense that would not result in any bullets in the marketing brochures.  Usability does not sell licenses, features do.

377

Uml Process / Re: Show inherited connectors between classes

« on: March 28, 2019, 02:00:22 pm »

Hmmmm Hmmm, what's a connector between classes?  If we are talking about an association - I am under the impression that the UML tool should not recreate the parent association for all its children - for example of the parent is removed from the diagram the association should disappear from the diagram too.

378

General Board / Re: EA not updating Foreign Key when the PK is changed?

« on: March 19, 2019, 12:14:39 pm »

As Paolo says... 2nd case.  Changes in PK of parent entity should be automatically reflected in the child entity FK.  If they are not, then the child table FK is wrong.  I guess EA won't be my data modeling solution.

379

General Board / EA not updating Foreign Key when the PK is changed?

« on: March 19, 2019, 01:39:08 am »

Using Data Modeling MDG it seems that if I change the PK name or I change the parent PK to another attribute, nothing happens to the child FK... it just retains the attribute and datatype of when the relationship was first created.  I have to change the FK manually using the relationship editor.  Is this normal or is there a better way?  EA 13 professional... don't have the (expensive) data modeling version of EA (I don't know if it would make any difference?). 
If FKs do not auto update, I'll try something else for data modeling.  Thanks for any advice!

380

General Board / Re: Why insert classes below components?

« on: March 08, 2019, 02:17:16 am »

I'm pretty sure that in UML 2.5 a component is just a class.  But I'm too lazy to actually look it up.

381

Uml Process / Re: Organizing Use Cases

« on: March 02, 2019, 05:57:32 pm »

Here's my perspective
yes you can get down this rabbit hole but there's a better way.
You are mixing what the user wants to achieve with how the system does it.  This is the typical pitfall of functional decomposition.  Very non-UML and very non-object oriented.  It leads to very large, complicated and less useful models. 

My approach would be: neither is a good use case.  The use case should show WHO wants to check authorization and object existence and WHY they want to check it.  In other words, what are they trying to achieve?

The mechanics of HOW the system achieves the user goals should be completely out of the use case model.  The mechanics should be described in sequence diagrams, because the lifelines will guide you directly to define a good, simple and efficient class diagram. 

Activity diagrams are of very little value because they show procedural logic without reference to the objects or their classes.  So you don't get much benefit from them.  Activity diagrams are useful to define class methods and user interfaces.  If you need to describe complex object transformations, statecharts are much better.

End of rant: use case models (and package diagrams) are deceptively simple and easy to draw.  But it takes a few years of practice to use them correctly (system architecture, implementation approach, project management, vendor management, testing....).  Just keep practicing and eventually you'll get it!

Hello,

I've been taking a crack at loading use cases into Sparx and ran into a situation where I'd like to get guidance.

I have a number of use cases that rely on a common set of steps.  Namely: checking if an object exists and checking if the requester has access to said object.  I'd like to factor these steps out into their own standalone use cases so that I can include their behavior when necessary.

As a result, I created 2 use cases - "Check Authorization" and "Check Object Exists".  They are each self-contained and have happy/sad paths as you would expect.  In my primary use case, I "included" the auth check and exists check into the scenario steps where I expect them.  All seems reasonable, however...

If I try to generate test cases or activity diagrams, this inclusion is lost. Activity diagrams no longer show branching that in the primary use case.  Test cases don't generate variation.

My question is if the use of inclusion is recommended in this situation?  I'm really just trying to avoid repeating alternate flow definitions and I want all the goodness that the tool can offer.  How have you managed this kind of setup


Thanks,

Alejandro

382

Automation Interface, Add-Ins and Tools / Re: 2 stereotypes on attribute

« on: March 01, 2019, 08:31:53 pm »

Eve and Qwerty, thank you and I look forward to learning more if anyone has further insights

383

Automation Interface, Add-Ins and Tools / Re: 2 stereotypes on attribute

« on: March 01, 2019, 03:24:54 pm »

Per UML 2.5 can an element have more than one stereotype?  Asking for a friend.

384

General Board / Re: ETL diagrams

« on: February 12, 2019, 12:05:35 pm »

How about using UML sequence diagram. Since ETL process is data transformation.  The sequence diagram can show the extraction from source data, transformations performed, and new data creation ready for loading.

385

Uml Process / Re: Data Modelling - how to flatten class inheritance?

« on: February 06, 2019, 04:17:01 pm »

Hi, I mean you have to manually edit the model by adding / deleting tables, columns and keys.

Yes I remember also ERWin had something like this but keep in mind that EA is much cheaper and just an excellent UML tool with a nice extension to support data modeling.  It is not a fully fledged data modeling tool nor it aspires / claims to be.

386

Uml Process / Re: Data Modelling - how to flatten class inheritance?

« on: February 06, 2019, 12:33:00 am »

Hello DS I think the most common way is to do it yourself by hand.

It makes sense for this not to be automated because you as data modeler have different choices on how to translate generalization/specification (at logical level) into physical tables.

You can google this but basically you could merge the child classes with parent into one big table; or replicate the parent attributes/relationships to each of the children (along with any other database objects: indexes, triggers etc.); or set up a parent class with identifying relationship to its children.  Etc.  You have to pick the right solution based on your physical data model design criteria.  Good luck!

387

Automation Interface, Add-Ins and Tools / Re: How to make references between attributes?

« on: January 31, 2019, 01:44:36 pm »

Svend can you elaborate on your goal?

If it's a class diagram, public attributes are public and private attributes are private.  In any case it's the operations that can change the attribute values.

If it's a data model in UML, if it's at conceptual level it should have very few or no attributes; and if it's logical it should be normalized (attributes depend entirely and only on the table-class primary key)

What is the scenario where you would need to include dependencies between attributes in your model?

388

Bugs and Issues / Re: requirement, issue, change etc elements in a sequence diagram

« on: January 23, 2019, 01:09:23 pm »

My understanding is:  if you drag a classifier on a sequence diagram, if will be represented as a lifeline!  So I guess the items you refer to are classifiers as far as EA is concerned.  Suggest to use text boxes and notes instead.
I can annotate sequence diagrams with very little problem, the only issue is that the autorouting of the note links is a little wonky and need to manually route (EA 13).

389

General Board / Information engineering connectors don't display aggregation and generalization!

« on: January 19, 2019, 01:32:05 pm »

I have a UML class diagram (conceptual level) that I want to show as conceptual ERD (using crow's feet notation = Information Engineering connectors in the diagram properties options) for UML-allergic stakeholders.

The diagram properties toggle from UML to IE works fine but generalization and aggregation connectors are shown with UML notation while everything else is changed to crow's feet.  Aggregation should be changed to a non-identifying 1:m relationship and generalization should be changed to the ERD the half-circle notation (a' la ERWin)

To me it seems that the IE connector implementation in EA is incomplete.  Is there a workaround?  Or should open a bug report?  Thanks
(EA 13.0.1310)

390

General Board / Re: Importing DB Schema From ODBC error

« on: January 10, 2019, 05:14:27 pm »

Not being sarcastic here, just realistic.
It may just be the case that your team could benefit working with an experienced contractor or 3rd party who can show you the way.  Several on this forum.  Good luck!