The Model Driven Software Network

Raise your level of abstraction

Just a short note.


Programming languages do a 2-way abstraction:

1. from machine level 'upwards', leaving out the technical bits.
2. from (all addressed) problem domains 'downwards', promising that the language constructs provided correspond to the most important real world structures.

In discussions on programming and abstraction, the latter seems a bit underrepresented, although not less important.  

 

That's why I just wanted to mention it here briefly.

|=

Views: 349

Add a Comment

You need to be a member of The Model Driven Software Network to add comments!

Join The Model Driven Software Network

Comment by Andreas Leue on September 28, 2011 at 7:55

sorry, "respective"

 

Comment by Mountriver TY Yu on September 28, 2011 at 3:53
the repective metamodel?
Comment by Andreas Leue on September 27, 2011 at 7:31

Yes, for a given language, that should hold: metamodel M3 (M2, M1) is typically more neutral than corresponding models M2 (M1, M0), otherwise the design of the repective metamodel is to be questioned.

 

Comment by Mountriver TY Yu on September 27, 2011 at 2:25

Thank Andreas for expounding!

I think I got some more understandings to your 'neutral'...

It's perhaps can be said that, a meta-model is usually more neutral, even complete neutral...

Comment by Andreas Leue on September 26, 2011 at 16:55

As for neutrality one can distinguish different abstractions; on the domain side there is

- independence of the model from some domain aspects (like e.g. finding commonalities between businesses)

- independence of the meta model from domains (going from DSL to something like UML)

- independence of all model levels from platform stuff (opposite direction, very desirable)

 and on the technical side

- independence of the model from all domains (just kidding, would be rather useless)

- independence of the meta model from all domains (desirable, increases reusability)

- independence from technical details (going from something like UML down to code)

 Depending on the kind of abstraction, even complete neutrality is possible (e.g. independence of metamodel from domains and independence of meta model and model from platform details)

Comment by Mountriver TY Yu on September 26, 2011 at 16:18

I can't make a clear conclusion to the 'neutral' yet.

Perhaps we might well talk about 'models' instead of the 'abstractions'.

There are two models from two objects (the source the model be abstracted from): the domain model(s) and the system model(s). In my recent thoughts, a black-box model of the app system will belonged both the system domain and the applied domain. Is this a neutral one?

Comment by Model Practice on September 26, 2011 at 15:30
Hi TY

"... they have different own source or basis to abstract"

so, do different abstraction types make it generally impossible to determine a reasonable >most< neutral position?

|=
Comment by Mountriver TY Yu on September 26, 2011 at 14:37
I think the key point is two ways, IMHO, they have different own source or basis to abstract.. (some related discussions on my blog at here)
Comment by Mike Finn on September 25, 2011 at 14:48

Abstraction only goes one way - upwards.

 

Comment by Model Practice on September 25, 2011 at 13:31
Hi altogether

think the "most neutral position" expression is one that I'm going to adopt, where I find the "most" (what makes it an optimisation problem) is a really challenging bit.

Have fun
|=

Badge

Loading…

© 2019   Created by Mark Dalgarno.   Powered by

Badges  |  Report an Issue  |  Terms of Service