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 - Geert Bellekens

Pages: 1 ... 5 6 [7] 8 9 ... 559
General Board / Re: use case non human actor
« on: August 28, 2018, 05:18:52 pm »
"Actor with a blindfold" is not a standard UML icon and AFAIK is not available in EA.

You can however always set an alternate appearance in EA, or use a shapescript to create this icon.


I still think you are better off keeping GUID's.

One (of the many) reasons:
What if you accidentally destroy a part of the model?
If you kept the same GUID's you can still reconstruct it from a backup.

About the GUID's in EA: Yes in most places they are regular GUID's.
Just a few places (like tagged values) have some weird repeating sections in their GUID's.


Great. That helps define our options. Since it's designed to be used by EA to generate the standard XSD, it sounds like it wouldn't be the best way to go. We need the output XSDs to have some significant customization. Starting from scratch would probably make more sense. Would you agree? Any kind of ballpark range for how much this would cost to develop?

Well, I'm not quite sure. It depends on the type of customization you need.
But, I did find that writing a tool to generate a specific style of XSD from a specific style of EA model is surprisingly easy/fast, compared to writing a generic XSD generator for any style of model.
(I recently did a rewrite of the Vienna add-in for a customer to generate XSD's from a UMM/UN-CEFACT model)

Ballpark estimate: something between 7 and 15 days of development.
That is of course still a serious investment, so it might be cheaper to go the post-processing road. Big advantage of the custom development vs the post processing is that you have 100% control and freedom and you are not dependent on the results of the standard XSD generation from EA.



Geert - as I am not 100% on what I am looking for & there is a LOT of shapes in your library, is there any particular shape that you might recommend as a  good example?
I can't really. Only a few were written by me. The others were extracted from existing the Sparx MDG technologies.


Hi Nick,

You might want to use a different image hosting site like

The one you are using is not exactly safe for work. A bit annoying for those of us using the forum at work  ;)



I don't know that specific one, but you can usually figure it out by checking the database.
If you know where EA stores it in the database then you can often figure out how to access it using the API.
Sometimes exotic things like this are not directly exposes in the API. In that case you can use Repository.SQLQuery to get the info from the database directly.

If you have a model on SQL server then the SQL profiler can be a very useful tool in figuring out where EA stores stuff.

If I would need to guess I would bet that you will find it in t_xref and that it is not exposed by the API.


So then it's safe to say I could not use the EA Message Composer to customize the XSD further to add text based on some Tagged Values, correct? Do you think that is something that could be accomplished by modifying the EA Message Composer source code, and do you offer that option? Since you didn't mention it as an option, I assume there are no MDG Technologies for something like this.
Yes. The EA Message composer is basically a tool to create a subset of a model, with some added intelligence. Think creating message definitions from a large canonical model.
It it not specifically aimed at XSD models although it does have some features that are specific to the XSD generation mechanism.
You could probably change the code to make it do what you need (or pay me to do it). You would need to evaluate if this would be the right direction given the current focus of the tool.

Also, does the EA Message Composer use the TPos field to set the ordering explicitly? Based on your message I'm not clear as to whether the TPos value is used in the generation of the XSD.

It uses the pos field of the attributes, and the position tag of the source connector ends for associations to play with the sorting.
Not sure about the order of the classes themselves. I don't think tt does anything specific with that, but I would need to check the code to be sure.


Bugs and Issues / Re: EA Connector for JIRA and quotation marks.
« on: August 27, 2018, 11:03:54 pm »
You can always send in a feature request to automatically change regular quotes into depending on some kind of setting?
It should not be based on the language of Windows, or the regional settings, because that will cause issues for other users who want to keep the regular quotes.



The position in the project browser (pos) depends on a number of things.
If it's 0 then that means it has no (user) specified order and uses the standard EA ordering logic. (Which is a bit ordering by type, and a bit of alphabetic)

The EA Message Composer add-in creates a duplicate of a (part of a) model and sets the ordering explicitly alphabetically (depending on which options you enable in the settings)
From that duplicate model you can then generate XSD using the standard XSD generation.


I would do

- Create DBMS repository
- Transfer the largest .eap file to new DBMS repository
- XMI export the part you need from the second .eap repository
- XMI import into the DBMS repository.

XMI export keeps existing GUID's, which can be an advantage in a lot of scenario's.
I'm not sure if Copy/Past package does the same.


BPMN 2.0 is pretty clear on when to use what.

A business process object can (should) be used for any process from the most high level to the smallest ones. There is no size restriction to a business process object.
It always contains one or more steps that are displayed on the diagram owned by this business process.

The only "sort of" exception is a sub-process.
A sub-process is in fact nothing more that a named group of steps in your Business Process.
A sub-process is only known to the owning Business Process and cannot be called from another process => they are not re-usable.

As soon as you want to re-use a (sub)process you have to create a proper Business Process and call it using a call-activity.


PS. A business process object should never be used on a BPMN business process diagram. Always use a call-activity.
Only on "overview" diagrams that simply list all available business processes you can use business process elements.

Hi Geert,
Can you elaborate on item (3)? Are you referring to post-processing within EA or something external? Is there any ability to modify and customize the generation of xsd's from within EA? Is there a way to somehow integrate scripts within EA to different EA export features?

We're trying to replace an existing post-processing tool with a method that is internal to EA.



We have solved this (for different customers) in different ways:

- 1) only use attributes and no associations (limits some constructs such as anonymous choices)

- 2 ) use an add-in I wrote: EA Message Composer to select the message as a subset. The add-in takes care of the alphabetic order.

- 3) write a post-processing tool/script to order the resulting xsd's alphabetically.

Hi Bill,

I was referring to an external tool operating on the resulting XSD's. Depending on what tools/technologies you are comfortable with I guess this could be a script of some kind, an xslt, a small C# program etc...
AFAIK there is no way to customize the standard XSD generation.


General Board / Re: Protobuf .proto generation from a class diagram
« on: August 27, 2018, 06:48:55 pm »
Never heard of it.

I'm pretty sure it doesn't exist.


Pages: 1 ... 5 6 [7] 8 9 ... 559