The ORM Foundation

My company does research into ORM. At the moment we’re considering the virtues of an extension to ORM.

Of course, doing so (creating an extension to ORM) is no small feat, because the likelihood of counter-example and inconsistency must be taken into very careful consideration; as does the notion of working with the rest of the community to bring benefit and consensus to the majority.

I find it funny and very interesting that we live in a time where there is still consideration of extensions to ORM. ORM v2 is not a few years old, and ORM itself only decades.

The axioms of ORM, I would have thought, would have all been settled by now, but even at last year's (2007) conference, further extensions were considered.

It’s hard to know when things will settle down. For my money, I peg ORM to be here in 2000 years, in much the same way as Euclid’s Geometry, and because of the atomic nature of the methodology. With the very axiomatic nature of ORM, I see many of the axioms as being rock solid, and fit to stand the test of time.

What prompted me to write this, is that it is almost certain that new ways of using ORM will be found, and in as much as we have ‘Non-Euclidian Geometry’, there will be ‘Non-Halpinian ORM’. I find that very profound and scary in a way, because there will be debate. But it will be the ‘usefulness’ of variants and extensions to ORM that will prove their worth.

I don’t know what those uses/extensions are yet (we consider only ‘one’), but it will be great to watch it unfold. I’m really interested in what other people see, and what they think it means for the ‘common meta-model’ as it unfolds.

Hi Victor,

Very interesting thoughts on ORM and Euclidian geometry.  Glad you posted that here.

As it happens, I’ve been reading a book I picked up years ago (an English translation of a book written in the 20’s by Hans Reichenbach: The Philosophy of Space & Time).  There’s a lot of good stuff in there about visualization of Euclidian and non-Euclidian geometries.  While reading, I sometimes noticed some parallels [are parallels axiomatic in non-Halpian ORM?], between the two subjects.

May we all be around in 2000 years to see if you’re right.

BRN..

Hi John,

Thank you for contributing to this post. For my money, it'd be a shame if the philosophy of ORM wasn't discussed.

In envisaging 'parallels' and ORM, I believe that it is fair to say that we already have both (analogously) 'joining' and 'non-joining' in ORM, through the  concept of 'join subset constraints' (Halpi, Join Constraints, orm.net.).

If one considers an ORM model where Join Subset Constraints are not employed, a vast aray of values may happily exist side by side in seperate Entity Types (e.g. analogously in 'parallel').

But the moment a Join Subset Constraint is employed, that arrray is significantly diminished and determined by what's on the starting side of the constraint. Moreover...what's on the other side of the constraint (while ostensibly a 'superset') conforms to the will of what is in the 'subset' (i.e. the join forces a intersection (analogous with crossing 'lines') through the join and in a manner similar to Schrödinger's cat (the join part of the constraint only takes into effect when you 'query' the model, not when you populate it).

 I for the life of me couldn't accept this at first. And it ties into Ken Evan's extended research into whether ORM may be used 'descriptively' or 'to design'. Join Subset Constraints fall into the latter in my thinking.

At first we thought there was cause to keep the notion of 'subset' without the 'join', but as Dr Halpin quite rightly pointed out to me, 'subsets' may be achieved by way of deft use of mandatory role constraints.i.e. it serves no purpose to consider the 'join subset constraint' without both concepts in place.
For anyone following this thread, we no longer consider an extension to ORM in this light, and thank Dr Halpin for his valuable input.

So we didn't have to wait long lol, but I agree...may we all be around in 2000 years to see further more interesting developments in ORM.

Best regds
Victor

(For some reason all paragragh spacing were lost in my reply post - hope this serves as a more readable replacement)

I hope we’ll both have the time to discuss these matters further; and that more members here will take up your invitation to join in.  I can see that your (and a number of others members’), understanding of the logical framework for ORM as a FOL theory, (as well the practical implications), is more advanced than mine – but I’ve always been pretty good at catching up when it comes to concepts. 

Hi Brian, I don't know where I got John from. I apologise.

On Join Subset Constraint, I believe the intended interpretation, and we'll most likely find when it is implemented in NORMA, is that it operates in a manner that "If you can't query it, you may as well not be able to store it"....so my reference to Schrödinger's cat is only helpful/relevant in analysing how it works. It is the part where I say "it serves no purpose to consider the 'join subset constraint' without both concepts in place [subset & join]" that should be more acceptable to how the constraint works (i.e. restricting/constraining population sets)

On the whole I believe tools like InfoModeler didn't take off because they were hard/fiddly/timeconsuming to use, and I couldn't agree more that with your comments that ORM would be an appropriate 'front end' for a RDBMS. Of course, some users of NORMA may argue that that is exactly what they are using if for.

I think, don't be shy in creating a seperate threat about your question "a UofD to be considered as just the terms (linguistic molecules), of discussion about a domain; or is it the totality of terms, facts, concepts and perceptions appropriate to such a discussion?".

Let's ask Ken if we can have a thread just on Philosophy under Research, so we can discuss that there. This thread is for extentions to ORM.

I'll leave discussion on that to there, but as a heads up (and to stay relevant to this thread),....we live in a day and age where we live among the Euclids of this industry. Which is to say, it won't matter what anyone thinks about what we may or may not do in ORM, we may only go be the axioms and rules of inference of ORM, or venture to 'extend' ORM in some way.

Having said that, and if you've read any of my posts, I'm not beyond giving anything a good kick to see if it falls over. So I believe there is great merit in discussing the points of your suggestion.

Best regds
Vic