Re: A database theory resource - ideas

On Mar 16, 4:05 pm, "Cimode" <cim..._at_hotmail.com> wrote:
> On Mar 16, 3:58 pm, "JOG" <j..._at_cs.nott.ac.uk> wrote:
>
> > I found one of the most interesting parts in a gargantuan recent
> > thread was the listing of priorities for reaching a certain audience
> > and determining what one wants to get across (Thanks to bob for that.
> > You may object to his style but I've found there can be lot of value
> > in his posts).
>
> > Anyhow I have discovered a lot over the last year since my arrival,
> > and as such I often feel the urge to contribute back to general
> > knowledge in the field. So with the loss of sites such as dbdebunk and
> > the general lack of a simple central resource for database theory I am
> > intending to put together a form of FAQ site. Hopefully this will be
> > useful to reducing retro-activity in the field, as well as being a
> > useful educational resource. Additionally any audience is self-
> > selecting and this can only ease a lot of the frustrations I have in
> > trying to convince those in entrenched positions of advancements over
> > the last few decades. Perhaps it may even provide a reference link for
> > common arguments that continually arise. Who knows.
>
> Good idea.
>
> > Anyhow I am initially aiming to concisely cover the following topics:
>
> > * The vital nature of separating conceptual/logical/physical layers in
> > data modelling
>
> Agreed. Suggestion: Why not create a *Data Modeling* section to put
> the separation between logical and physical in perspective?
> Demonstrating that is perfectly applyable even using crappy SQL
> Systems is a strong argument against nonsense. On the same variation:
> why not distinguish an *Implementation* vs *Fundamental*
> perspective...Such structure may help sharpen the presentation of
> concepts...
>
> > * Working in terms of Propositions and not Objects
>
> Interesting. That would interesting for OO audiences desiring to
> learn more about data management...Maybe a section *RM for OO Thinking
> Developpers*
>
> > * The argument against OID's (and hence for the information principle)
>
> I am not convinced a defensive attitude toward OID aberration is
> effective in promoting RM. But maybe you could create a *Debunk* or
> *BS* section and put some strong critical articles...
>
> > * Why Navigation was replaced by Declaration
>
> Agreed. I would see it in some kind of *Why RM is superior*
>
> > * That Data models involve not just structure, but also manipulation
> > and integrity.
>
> Agreed.
>
> > * Why hypertext models are insufficient (due to irreducible tuples)
> > (web 3.0 ... good grief)
>
> Interesting.
>
> > * What semistructured data is (or rather is not).
>
> Agreed.
>
> > I am planning to omit well covered ground such as eliminating
> > redundancy and anomalies through normalization, simply referring to
> > external links. Obviously all of the above has been covered somewhere
> > in the literature, so the aim is rather to produce a central, concise
> > and hence accessible resource as opposed to resorting to a text-book/
> > academic paper format.
>
> Good luck. I encourage you in doing that.
>
> > The main purpose is to provide a purely educational resource, with as
> > little impartiality as I can muster, constraining to established
> > theory and facts, or clear logical arguments.
>
> > So my question to cdt is to ask what /you/ believe the priorities for
> > such a resource would be?
> > - which pivotal questions are most misunderstood?
> > - where does most ignorance lie in our field?
> > - are there are any crucial topics that you believe it would be useful
> > to address that I have not listed.
>
> Some articles I wish I could see in a site...
>
> 1) The Impedance Mismatch Myth
> 2) On why are set operations are intrinsically superior to procedural
> approaches
> 3) On the limitations of direct image dbms's systems (ORACLE, DB2, SQL
> Server) as opposed to non direct image systems
> 4) On relating the current systems and architectures promoted by major
> editors and why they are doomed to fail
> ...and so forth...
>
> > Any input is gratefully received.
>
> See above...Let me know if you need any help (hosting).
>
> Hope this helps...

Hi cimode - thanks for the feedback. I have found your posts a lot more constructive recently. I would say that I don't want the resource to seem like an RM advert, rather something more neutral, focusing on the theory of what makes /good data modelling/. It just so happens that RM matches most of them, but I don't want to preclude that there will be other models in the future that also do (I don't think any of us think the current RM will be the last word). I've summarized some of your suggestions:

• A data modelling introduction section (good idea)
• Talk about the positive nature of content-based addressing as opposed to just the negative aspects of OID's.
• On why are set operations are preferable to procedural approaches

I am currently uncertain about "The Impedance Mismatch Myth", and your other suggestions are outside of my remit of knowledge so I would not be confident in addressing them thoroughly. regards J. Received on Sat Mar 17 2007 - 16:16:34 CET