Alternate link endpoints based on mulitiplicity

[kgish]

Is it possible to define other associations endpoints depending on multiplicities?

I'm thinking about the single, optional, optional-to-many and one-to-many relationships traditional to text-book database diagrams:

+----- for 1

o----- for 0,1

>+---- for 1..*

>o---- for 0..*

Is this possible? (sorry if the schematic drawings above look ugly, but hopefully you know what I mean).

[Paolo F Cantoni]

The answer will be no...  We Data Modellers are looked down on by the Objectors and ALL UML modelling tools are built by OO programmers...

Your chances of getting this chaged are very slight to zero.  But submit a formal enhacement request anyway...

Paolo (Data Architect)

BTW: Interestingly enough, I have NEVER yet found an OO developed application that could withstand a normalized Database designed with proper referential integrity...  They NEVER get it right! (At least for any application of other than trivial complexity).
One recent use was to use the "correct" DB to track down intermittent errors in the application code by tracking down, identifying and analysing data errors in the production DB designed by programmers - by using the referentially integral DB as the "Gold Standard".

[AshK]

Gentlemen,

Excellent suggestion; coincidently supporting this type of notation is quite high on our todo list.

Our intention is to help data modellers such as yourselves feel more comfortable in EA.

I can give an ETA of around 1 month - so, please expect to see this expansion within a build or two.

Best regards,

Ash

[kgish]

This change would be a GREAT improvement, as it would result in daigrams that are more readable for those non-UML manager-like folks. Thanks alot!

Would it possible to be notified someway, e.g. via email.

[Paolo F Cantoni]

Gentlemen,

Excellent suggestion; coincidently supporting this type of notation is quite high on our todo list.

Our intention is to help data modellers such as yourselves feel more comfortable in EA.

I can give an ETA of around 1 month - so, please expect to see this expansion within a build or two.

Best regards,

Ash

Will Sparx be supporting both "standard" ERA symbols like kgish mentioned and the IDEF1X notation (extensively used in the US)?

If you were feeling really ambitious you could also include the Oracle notation (I think developed by Barker?).

Paolo

For once, I'm really happy to be proved wrong!

[kgish]

Nothing bad about being proven wrong (once in a while).

[michael_rainwater]

Sweet!

[AshK]

Will Sparx be supporting both "standard" ERA symbols like kgish mentioned and the IDEF1X notation (extensively used in the US)?

If you were feeling really ambitious you could also include the Oracle notation (I think developed by Barker?).

Your suggestion is most appreciated Paolo; at this stage, we will be adding support for the connection endpoint styles of Barker's ER modelling.  

Other data modelling notations such as IDEF1X and Oracle will definitely be considered for future expansions.

Ash.

[AshK]

This change would be a GREAT improvement, as it would result in daigrams that are more readable for those non-UML manager-like folks. Thanks alot!

Would it possible to be notified someway, e.g. via email.

Sure thing, I'd be happy to email you (aswell as updating this thread).

Also, many thanks to all for the positive feedback.  

Ash

[sargasso]

Whoa there pardners!

Barker is not exactly equal to Oracle.  See Dave Hay's site http://www.essentialstrategies.com/index.htm for "A Comparison of Data Modeling Techniques" in his publications for the variants.  You may need to register (for which you will get a lovely postcard).

UPDATE: Sorry - that is not the one that shows the barker / Oracle variants.

My vote, for what it counts, is for IDEF1X, then IE, then Barker  (but I will probably continue to use my favourite $big tool for ER modelling anyway - I reckon EA should stick to UML).


bruce