Baseline feature

[ToniPepperoni]

Hello guys,

I am a very satisfied User of the Enterprise Architect (Version 9.3) and I use this Tool in various projects in our company. Since the projects are going to be more and more complex I decided to develop a “Best Practice” for our projects.
For Example, I want to use the baseline feature in the following way:
Every System, which is developed in our department, has a package under the root-model. The packages (and so the systems too) have some dependencies. Every package has also its own baseline, which is mirrored in its own system-version.

Since the projects have some dependencies there are problems by using this in the above-mentioned way. For example, system X in Version 1.1 have a dependency with system Y in version 5.4, but not in version 5.5 and 5.3.

How can I ensure that my model is semantically correct? My current “solution” is to create a new baseline for the root model and every time I create a new baseline for a system I create another one for the root. But this solution is very problematic, because I am not the only one who is modeling and what happens when I want to edit a baseline and overwrite the baseline (by deleting the old one)?

All the sources I read in the web or in the printed guide “Projektabwicklung mit UML und Enterprise Architect” explained the baseline-feature in a “Hello-World”-way and not in my specific way.

Informations:
OS: Win7 32-Bit
EA: Version: 9.3.935
Ultimate Edition

I hope you can help me.

Best Regards,
ToniPepperoni

Edit: Please try to ignore my language shortcomings :-)

[qwerty]

What a coincidence, but yesterday I asked someone else to first use the search button top right (extend search time for a year) and look for "baseline".

q.

[Helmut Ortmann]

Hello,

I would consider using Version Control. Then you can make baselines with your Version Control Tool.

Helmut

[ToniPepperoni]

My german is much better than my english, so I will go in german now.

Hallo Helmut,
zunächst einmal danke für deine schnelle Antwort. Ich glaube, ich habe mich nicht richtig ausdrücken können. Mein Problem ist nicht die Versionskontrolle an sich, sondern wie ich es am geschicktesten manage. Ich möchte den Enterprise Architect dahin gehend nutzen, dass ich nicht nur den momentanen Schnappschuss meiner Systemlandschaft modelliere, sondern auch die Entwicklung. Ich möchte mitmodellieren, wie sich jedes einzelne System und die Systemlandschaft als Ganzes entwickelt hat. Daher wollte ich mit den Baselines die Versionen der Systeme widerspiegeln

(Beispiel: System X hat in der Realität eine Version 12.1 und dementsprechend hat sein Package im Modell ebenfalls die Version 12.1. Wird ein neues Release gepublisht, so modelliere ich die Änderungen gegebenfalls im Modell und ändere die Version des packages).

So hätte ich alle Entscheidungen während der Architekturmodellierung /-entwicklung dokumentiert.

Ich hoffe, ich konnte verständlich machen, auf was ich genau hinaus will.

Das Problem ist nun, wie halte ich die Gesamtmodellierung mit den verschiedenen Systemen konsistent (wie im Initial-Post beschrieben System X in Version 1.1 hat eine Schnittstelle mit System Y in Version 5.4 aber keine zu Version 5.3 oder 5.5)?

Grüße,
ToniPepperoni

[qwerty]

Hallo Toni,
(sorry; I'll add an essence of the thread later, if there's more than the forum already covers)
meine Antwort ist damit schon auf deine Frage eingegangen. Hier ist schon einiges zum Thema Baseline hinterlegt, was du lesen solltest. Um er kurz zu machen: Baseline ist kein Konzept, das man auf UML-Modelle anwenden kann. Das klappt für Quelltext, aber nicht bei UML. Dort ist eine einzelne Änderung nicht lokal begrenzt sondern hat umfassende Auswirkung. Meine Empfehlung: nächtlicher Snapshot (und einchecken in ein VC System) und ein zentraler Repos mit Require Lock to Edit.

q.

[ToniPepperoni]

Well, I guess you did not really understand what I want and what my problem is :-)
I do know how to use VC in Enterprise Architect and how the baseline-feature technically works, but I do not want to use it as a simple repository (e.g. code-repositories) for distributed team work. I want to use it to express how the systems have developed by each version. So I do not want to have tons of baselines (for each change!) , I only want a baseline for each Version (as I try to told you system X has version 1.1, so the baseline of the package too and so on).

Why do I want to use it? Well, there are some use-cases/advantages for that. For example, I could model the future version of system x to show my management or my developers how the systems should look in the future. Another advantage is, that I just can see the main differences of a systems.

I do not know whether the baselines are the solution for what I want, but my intention to wrote here was that I might be not the only one who want the same. So thats why I hoped for some experiences and best practices. I guess a board is exactly that, a place to exchange experiences.

Und wenn du keine andere Antwort hast, als Leute auf Posts aus der Steinzeit zu verweisen, dann antworte bitte nicht! Denn es ist unverschämt mir zu unterstellen, ich hätte weder die Suche benutzt noch anderweitige Artikel gelesen!!

[qwerty]

Und wenn du keine andere Antwort hast, als Leute auf Posts aus der Steinzeit zu verweisen, dann antworte bitte nicht! Denn es ist unverschämt mir zu unterstellen, ich hätte weder die Suche benutzt noch anderweitige Artikel gelesen!!

[translation]If you don't have another answer than pointing others to posts from stone age, then please do not answer. It is impertinent to assume I had not used search nor read other articles![/translation]

Sure. Every newbie entering here has always used search and read all relevant posts. How could I have missed that. I'm soooo sorry.

q.

[Stefan Bolleininger]

At First:

You can baseline a model   - it is depending on your modelling "dialect" if it is useful. In my cases.. it is  

@Toni: Please tell me, if I'm wrong :
Your model is a collection of systems which can interact.. but not in all circumstances.
All Systems within the model are developing independent of the others and could require cross references.
The "product" of each System can be standalone (with its dependencies) or can be a full suite.

Your current problem:
You cannot verify in each baseline that every dependency is fullfilled by the other systems, or if the dependency has been deleted.

Is my assumption right?

Regards

Stefan

[ToniPepperoni]

You nailed it!

Another problem is that I have cross-system AND cross-baseline dependencies. I tried it in the follwing way:

The package structure
1. My Model (own version 2.1)
    (1.1. some cross-system views --> version like "My Model")
     1.2. System 1 (own version 6.4)
     1.3. System 2 (own version 1.1)
      .....
     1.4. System X (own version x.x.)

The "My Model" version is always a snapshopt from the whole "systemlandscape", for example: version 2.1. of "My Model" contains System 1 v6.4, System 2 v1.1, but not system x, because this system is developed later in "My Model" version 3.1...
Now I realised that I forgot to model something in System 1 v6.4 (which is a part of "My Model" v2.1.), so I want to edit the baseline and now you can see that this way brings some inconsistency, because I have to edit the System baseline AND the "My Model" baseline.

I hope I could express my problem better with these example  :-/

So is there any best practice for implementing this? The headline of this topic is maybe not good, because I dont know whether the basline is the best solution, there are maybe better mechanism in EA.

Regards,
Toni

Edit: Correcting some misspellings

[Stefan Bolleininger]

Hi Toni,

since your complete System is just a compilation(packaging) of the subsystems, I would recommend you not to use the model, but crating a own node for it.
Than you create your "deployment list/packaging model - or how you want to call it"
This part can be baselines independent from all others and only reflects the current systemcompilation.

Regards

Stefan