Re: Relational Databases Lack Relationships

On 2015-10-28 07:05:07 +0000, Norbert_Paul said:

> Eric wrote:
>> On 2015-10-27, Norbert_Paul<norbertpauls_spambin_at_yahoo.com>  wrote:
>>> Nicola wrote:

>>> 

>>> 
>>> Commonplace but maybe wrong:
>>> http://comjnl.oxfordjournals.org/content/53/1/69
>> 
>> Oh, we knew it was wrong, but thank you for the link to help prove it!
> Oh, I didn't know you knew it. Your post made me think you shared
> Egenhofer's point of view.
> You are welcome.

I have followed (distractedly, though) your interesting discussion in this group about your topological model a few months ago, but I have not read your paper (btw, just a few days ago I was searching for your Lisp implementation, but I could not find it).

As for my point of view, I do not share the opinion that the relational model is not expressive enough to capture, say, spatial data. On the other hand, I think that there is a tension between what you choose to represent as a data type and what you represent as a relational schema, which is determined not only by expressiveness, but also by the succintness of the representation, by the desired level of granularity of the representation, and by the data integrity requirements. For some kinds of data, it may be legitimately, and perhaps convincingly, argued that the tension resolves favorably on the side of using specialized data types. As an aside, having worked with both spatial and genetic data, I find it curious that for the latter no one, as far as I know, has argued in that sense, which in my opinion would be more justified than for the former.

Besides the above considerations, which are purely at the logical level, there are physical issues as well. Currently, the most common implementation of a relation in a DBMS is a row-ordered record-based file (with a one-one mapping between a table definition and a file). Saying that some types of data are not efficiently managed when stored in a row-ordered record-based file (which is often what is really meant by the misleading "data do not fit into the relational model") is an unproblematic statement.

Nicola

• news://freenews.netfront.net/ - complaints: news_at_netfront.net ---