1 |
On Monday 25 October 2004 10:07, Maurice van der Pot wrote: |
2 |
> |
3 |
> So in your view, what Ed and I are suggesting is more complex than the |
4 |
> original proposal? I definitely disagree on that. The implementation of |
5 |
> what I proposed is simple to implement (handle a group at a time, |
6 |
> automatically making circular references impossible), simple to debug |
7 |
> (groups do not influence eachother unless one lists the other, evaluation |
8 |
> is not order dependent) and simple to use (if a user doesn't think about it |
9 |
> a lot, it will still have the expected results in cases like @KDE -@GNOME). |
10 |
|
11 |
Basically I agree with you both. Maurice's solution is complicated, not |
12 |
straight-forward and probably complicated for people who don't get set theory |
13 |
all that well. |
14 |
|
15 |
However the proposed solution can also get awkward in certain cases. For |
16 |
example what does the negation of a group has as result when it contains a |
17 |
negated element. |
18 |
|
19 |
What I think is the best solution is to allow negation of groups only under |
20 |
certain conditions. One condition is that the group may not contain negations |
21 |
|
22 |
Paul |
23 |
|
24 |
-- |
25 |
Paul de Vrieze |
26 |
Gentoo Developer |
27 |
Mail: pauldv@g.o |
28 |
Homepage: http://www.devrieze.net |