Re: Objects and Relations

On Jan 31, 9:47 am, "Keith H Duggar" <dug..._at_alum.mit.edu> wrote:
> David BL wrote:
> > I found your post very informative and interesting and
> > shows me that RA is more powerful than I imagined - when
> > combined with aggregation. I was expecting that substring
> > testing would require transitive closure. I admit the RM
> > approach is a little foreign to me.
>
> "... imagined ... expecting ... foreign to me"
>
> There you admit that you have limited knowledge of and
> experience with RA/RM. Further, you admit that you were
> aware of your own ignorance (by itself a good thing). Yet
> all the while you threw out absolute (and wrong) claims
> such as:
>
> "Something important like searching for a given
> sub-string is actually quite difficult to specify
> in the RA."
>
> Does that seem dishonest or stupid to you?

No. I still think searching for a given sub-string is difficult in RA. I'm taking "difficult" to mean difficult to comprehend or come up with rather than "not concise". This is ultimately a subjective thing and therefore I apologise for not choosing my words more carefully. If anything Marshall's solution confirmed my original opinion. Given my lack of experience in RA I don't blame you for being dismissive of my opinion.

The fact is that I understand the operations for RA, understand set theory and FOL and also the ideas of aggregation. I don't consider myself particularly ignorant. IMO Marshall's solution is "tricky", at least compared to well written solutions expressed in procedural, functional or logic programming languages.

Perhaps I'm missing the mindset that comes with thinking in RA/RM, and you could suggest a suitable reference

I understand that I need to think in terms of set level manipulation. For example, to find the *second* match, procedurally I would simply ignore the first match. In RA I presume I would need to use a set difference to remove the first match. Is that right? Can this be written elegantly - ie without repeated sub-expressions? Furthermore is there some theory that maps this back to an efficient implementation?

> To me you seem quite intelligent; so don't act stupid. Take
> a chill pill and stop using words like "onus". Do that and
> you can probably learn much here and some of us (certainly
> including me) may also learn much from the process.

What usage of "onus" are you referring to? Received on Wed Jan 31 2007 - 03:59:52 CET