Failed SCEA - short by one point

[Tomasz Romanowski]

I got my score which shows I was short just by 1.6 points. I see I lost most of the points in the class diagram but the assessor did not say what was wrong.
I'm looking at my class diagram and I'm not quite sure what the problem is. I'm listing the controllers and the entities. I'm listing all the methods shown in the sequence diagram. I'm listing the major attributes. I'm showing the relationships amongst the entities. For the controllers most of the relationships are "uses".
I'm also showing how the controllers are used by JSPs.
Overall got about 9 entity classes and 8 controllers.
Any ideas why I lost so many marks?
Also, I've searched the forum for the resubmission process and from what I've read I'm resubmitting only part 2 right? Part 3 stays the same, with same amount of points? How much time to resubmit?

[talu singh]

Does your class diagram look similar to the CADE's - Sun Certified Enterprise Architect for Java EE Study Guide (2nd Edition) example?

Did you preserve the business domain model relationship when you created the class diagram.

[Tomasz Romanowski]

I did preserve the relationships but I did not read the book. Why is this supposed to be like a sure-fire pattern to follow?

[Ryan Fernandes]

I guess they mention in the assignment that if you fail, they will let you know which sections need improvement. You should contact them and request for this information. It will be important.

[Mihai Lihatchi]

Hello Tomasz.
Thank you for your post , I am sure you are upset with the result but posts like these help even more then the "I Passed" posts .
Please confirm with us what went wrong and the lessons learned in the process when you get feedback from Sun.
Between the Sun/Oracle feedback and your expertise you should get something together to pass the certification.

I am trying to get my assignment finished and to understand as much as possible from fellow candidates experience.
PS : I always learned more from my failures then from my successes.

[Tomasz Romanowski]

What exactly Sun found wrong remains to be seen but most likely it was my attempt to "improve" the data model and dismissing some of the classes as unnecessary replacing them with one generic class.
I'm going to resubmit based on that assumption.

[Kuppusamy Venkatasubramanian]

Hi,

Did you get any reply from Sun on which part of your assignment is having issue and what is the problem? It will be helpful for others to avoid those mistakes.

Cheers,
Kuppusamy.V.,

[Tomasz Romanowski]

Resubmitted and passed.

[Srinivasan Rengan]

Hey Congrats...That is the spirit buddy :-)

Now that you have had a roller coaster, you would be in a position to deliver a lecture to the world on a topic say "Dos and Don't for the exam" better than most of them :-) Assuming that it would seriously help a person like me who is not finding the entry door to the maze...

In short...what made the difference this time round?

Thanks
Srini

[deepak adlakha]

Congrats Tomasz Romanowski

[Arpit Kumar Jain]

Congrats Buddy

[Praison Selvaraj]

Congrats, Tomasz

Cheers!,
Praison

[Tomasz Romanowski]

Like I said earlier, what I did was nothing extraordinary, apparently it's a very common sin that mostly affects the class diagram. Basically I "improved" the domain model and threw away some of the classes replacing them with two generic ones. It affected the scores on my class diagram as well as assumptions/risks/mitigations list. Wen resubmitting I dutifully replicated all domain classes.
Basically for SCEA domain model is not to be f..d with. You just need to assume it's perfect.

[Mihai Lihatchi]

Thank you for your answer Tomasz .
Your feedback is worth a lot. I guess the part 3 essay was not affected by all this stuff ? I am trying to understand
how to get max points for my essay (because I will not redo it ).

Congratulations for passing the SCEA certification.

[Deepak Bala]

Tomasz Romanowski wrote:Resubmitted and passed.

Congrats Tomasz

[Tomasz Romanowski]

Thanks folks. Correct, you don't resubmit Part 3 (essay). There is relatively not much to gain from it since it's worth only 24 points.

[Rajiv Shr]

Tomasz Romanowski wrote:Like I said earlier, what I did was nothing extraordinary, apparently it's a very common sin that mostly affects the class diagram. Basically I "improved" the domain model and threw away some of the classes replacing them with two generic ones. It affected the scores on my class diagram as well as assumptions/risks/mitigations list. Wen resubmitting I dutifully replicated all domain classes.
Basically for SCEA domain model is not to be f..d with. You just need to assume it's perfect.

Hey, I am doing the same thing. Basically, in the SCEA domain model, there are different "parts", which I am replacing by a Part Superclass, whose children are PartCategory and a PartType. Part will be a base table, and PartCategory and PartType will be a List of value code tables in the database. In the assumptions, I am justifying this by saying that with this design, it is easy store different categories and type of parts, and also it will be easier to add new parts in the future.

Are you saying, we should not do this? We should not modify the domain model even one bit, even if it is to improve the design? (Note that I am not changing the relationships. Instead of several parts linked to a 'Request', I am having the superclass Part linked to the Request. Just that the different parts are replaced by 2 generic classes - PartCategory and PartType)

Further, what does everyone think of correction of multiplicity in the domain model? Example: The assigment says that Multiple Requests can be related to Multiple Parts; but this is impractical according to me. I am changing it to 1 Request can only be related to 1 Part. Are such corrections allowed?

Your statement that the domain model should not be f..d with, would not always be the best choice in a real-world scenario.

[Jeanne Boyarsky]

Rajiv Shr wrote:Are you saying, we should not do this? We should not modify the domain model even one bit, even if it is to improve the design?

Yes. I agree with this statement. Even the SCEA study guide recommends not changing it because you run the risk of them thinking yours is worse.

Rajiv Shr wrote:Your statement that the domain model should not be f..d with, would not always be the best choice in a real-world scenario.

This isn't the real world. This is a certification test.

[Kuppusamy Venkatasubramanian]

Hi,

In my case also, i had to change the BDM a bit as some of the relationships were not correct. I wrote that in the assumption as well as the reasoning for the assumption. In another case, though i maintained the multiplicity, during the class design, i assumed that the navigation is one way so, i didn't have to have a many-to-many bi-directional relationship which further simplied the diagram. Always ensure that whatever assumptions made in coming up with the diagram should be detailed in the assumption list as it will convey the basis of your design to the examiner (and it is the only descriptive way to convey whatever information you want to convey)..

All the best..

Kanpai..
Kuppusamy.V.,

[Rajiv Shr]

Also, in the Business Domain Model, they have shown an Entity A --> Entity B, and Entity B --> Entity C. Also, they show Entity A --> Entity C.
I feel A --> C link is unnecessary, as Entity C can get all the required information about Entity A from Entity B, since C holds an instance of B, and B holds an instance of A.

I hope we can make such corrections in the class diagram and provide justification in the assumption.

[Tomasz Romanowski]

Rajiv Shr wrote:
Are you saying, we should not do this? We should not modify the domain model even one bit, even if it is to improve the design? (Note that I am not changing the relationships. Instead of several parts linked to a 'Request', I am having the superclass Part linked to the Request. Just that the different parts are replaced by 2 generic classes - PartCategory and PartType)

This is just too funny. That's exactly what I did! I guess great minds think alike... Well, like I said, don't do that!

[Rajiv Shr]

Lol, but if we have all the different Part types as Entities, wouldnt that mean there is one table for each? I just didnt think anyone would want to model it like that. Imagine if there are 100s of different Part types. You would not want to have 100 tables in the database. Its just too inefficient.

Anyways, I just couldnt go with what was provided. Hopefully the examiner agrees with me