Book a Demo

Author Topic: Data modelling improvements  (Read 3922 times)

davidb

  • EA Novice
  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
Data modelling improvements
« on: September 06, 2006, 07:35:08 am »
Hi,

I’ve been using EA as a UML diagramming tool for the last 3 years and I like it a lot. Lately, we where looking for a database modeling tool. I suggested we gave EA a try to see how it can handle this job. Another project team is also doing a similar trial period with Visio. In the end we will decide which one will become the official database modelling tool

So for I have been very pleased with EA and it has many advantages over Visio (EA generally handles the graphics much better than Visio does). EA also has some weak points, most of them relate to productivity issues and the lack of some user defined options. I listed each case in a (personal) priority order below.

1. The creation of a table’s columns is more time-consuming than it should. It requires a lot of mouse clicks. That could be avoided by letting the user add and edit directly in the columns grid instead. The  column attributes that are the most commonly used are: name, type, size (where relevant), primary key and nullable. As a comparison, Visio is much faster when creating a lot of columns.
2. In the current version one has to manually create the foreign keys between to associated tables. Why not offer the possibility to “migrate” the PK of the target table to the source table if it is not present of associate them if they are present. Again, this would be a great productivity feature
3. A data model validation feature would be nice. It could help the designer spot missing elements. For example, it’s very hard to tell whether or not the FK have been set on any given relationship (especially if the association’s labels are hidden).
4. Let the user defines which prefix he wants to use for the foreign keys, primary keys, indexes, etc. For example, “UQ” is not a very common prefix and I have no way to change it to “UK” but to alter every unique key constraint (operation) in the model.
5. In the current version I had to select every association on every diagram individually and use the context menu to hide the labels. Instead, I’d like to be able to define how I want to view all the table associations in a global setting (setting it per diagram would be nice, but a global setting would be best).
6. It would be great to be able to define a column as an auto number in only one step. Also, the start value and the increment value could be both set to 1 by default (this value could even be user-defined).

Il also noted a bug while designing the data model
- If a UQ constraint (to “UK_XYZ” for example) in the operation form the name of the operation is changed in that window but not in the project explorer.

Thanks,

David

thomaskilian

  • Guest
Re: Data modelling improvements
« Reply #1 on: September 06, 2006, 07:54:40 am »
1. should also go for class operations too! Yes - please!
2.-6. I'm using a transformation to avoid these - I'm only 2nd degree data modeller.

Pls. don't forget to send also a bug report for that last topic.

jaxtrx

  • EA Novice
  • *
  • Posts: 1
  • Karma: +0/-0
  • I love YaBB 1G - SP1!
    • View Profile
Re: Data modelling improvements
« Reply #2 on: September 07, 2006, 05:23:29 am »
I'm having problems reverse engineering from a mysql 5.x database. The foreign keys show up in the table display, but are not associated with fields, and no association lines are drawn.

To reproduce, just right click on a datamodel, select import DDL via ODBC, select data source and I selected import as new objects.  (I also tried synchronize, but same problem)

Same technique works fine for Oracle.  I don't think it's a driver issue since the PK's & FK's do show up, more as index names, but do not get associated with the fields.

If required I can send schema.

Thanks in advance for speedy help on this!

Charles



Ivt

  • EA Novice
  • *
  • Posts: 1
  • Karma: +0/-0
  • I love YaBB 1G - SP1!
    • View Profile
Re: Data modelling improvements
« Reply #3 on: September 07, 2006, 06:27:28 am »
Hi,

Reverse engineering from Oracle only works up to a point. If entities/tables and their relations have been "reversed" into a datamodel diagram and (some of) the same entities have to be reversed into another datamodel diagram the relations don't show (foreign key constraints). I want to use submodels, but this feature/bug? makes that difficult. (Am using version 6.1 at present).
Also, when reverse engineering some new tables into an existing datamodel diagram, the layout changes that I have made are not preserved.
Fixes for both issues would be very much appreciated.

I.