Re: Constraints and Functional Dependencies

paul c wrote:
> ...
> Let me give a simplified small example relation that I hope will either
> show I don't understand TTM's GROUP or maybe show what I thought I meant:
> ...

Actually, I can probably give my own different example to show that I'm probably not following TTM's intent for GROUP:

R
{a}

---
{1
  2}
  1

(I'm trying to show that R has two rows, not three as the picture might 
suggest.)  If I UNGROUP {a}, I think I would like the result to contain 
the same two rows, ie., it would be as if UNGROUP had no effect!

My reason for this will seem vague until I can put it better, but 
basically I think R's meaning, whatever it is, is not the same if I give 
a result of:

R
a
-
1
2

It seems to me that we already have projection for this kind of 
existential qualification, so I don't see why UNGROUP should duplicate that.

(Another vague thought is that while I think an engine never knows the 
human meaning of a relation, it must do its darndest to not disturb all 
possible meanings.)

Also, I'm sure most will see all this as a distraction from Marshall's 
topic and I wouldn't argue with that.

p