Re: Question on Structuring Product Attributes

From: Derek Asirvadem <derek.asirvadem_at_gmail.com>
Date: Tue, 11 Oct 2011 15:44:45 -0700 (PDT)
Message-ID: <b4f9074a-86f4-4d72-8cc9-d8c00f7838d7_at_v18g2000pri.googlegroups.com>


Roy

To be clear, I am totally for:
- correct Normalisation, which results in Subtypes - and certainly a hierarchy, if that is what the exercise produces in any given situation.

- No Nullable columns, especially not FKs.
- simplicity, which is a classic result of the formal process.
- No "hiding", such that the model is plainly exposed in the catalogue

Credit for the concept or notion of Subtypes goes to Codd & Date; for the theoretical modelling, to Codd and Chen; and for practical modelling or implementation specifics, to R Brown.

I appreciate that SQL does not have formal support for Subtypes, as it does for eg. Declarative Referential Integrity, Foreign Keys (and even that was added years afterward). Which leads people to implementations such as this.

This method does have a bit of processing, and of hiding that behind Views, though.

My question was specifically regarding the method, of implementing RI for Subtypes. Who deserves credit for this technique ? (I am not suggesting plagiary here.) When I present this method to others, I would like to give due credit the original author, as we do when we discuss Joe's Nested Sets.

Yadda

> Is there a reference on the web or book that
> fleshes this concept out?  Thanks much.

(I assume you mean the concept of Subtypes, since Joe has provided the subtype RI method in fair detail.)

Well, no. Anyone with a keyboard can "publish" these days. The web is full of shallow information and misinformation; one has to dig deep through that morass to find nuggets of truth, and to do that one has to be able to discriminate between the two, which is not possible for the seeker. Also, there are many related issues within the one science; one aspect cannot be excised, held up, and inspected in isolation from the other aspects (which is the common problem afflicting any technical discussion on the web). All you need is a good text book and some formal study. Or work with a mentor over a period of time.

Subtypes have been with us since the 1980's, when the RM was first available in product form, and I and many others have used them since then.

Regards
Derek Received on Wed Oct 12 2011 - 00:44:45 CEST

Original text of this message