dhartford / allenintervalrelationships Goto Github PK

Your correction to fourth row:
IF
A di B
B d C
THEN
A ni C (using your notation of "ni" for all relationships)

Pictorial counterexample:
AAAAAAAA_________
__BBBB___________
_________CCCCCCCC

Or in other words; if A < C, then there is no possibility that premise B d C 
could be true.  Therefore, some scenarios simply can not exist and thus A ni C 
(which contains relationships A {<, >, m, mi} C) is a false conclusion.  

However, I do agree that Allen's notation is fuzzy when he says "o oi dur con 
=".  If he meant "o oi d di =", then why change notation to be different from 
the rest of the table?  I'm just guessing widly, but maybe he didn't have 
enough space to write "o oi d di s si f fi =" in one table cell.  I believe it 
should be {o, oi, d, di, s, si, f, fi, =} which you already picked up on in 
later rows.

Hi,

The current implementation of pathConsistency() is not able to know if a new 
constraint edge has been added or not, we need to go through new constraint 
edge one by one.

- What steps will reproduce the problem?
I attached the test case to reproduce the error output.

- What is the expected output? What do you see instead?
In the attached results.txt, we could see the output of the current 
implementation was the first matrix. However, the relation between 2 and 3 
should NOT be Full. Instead, the second matrix was what we expected.

I also attached two images for visualization.

Yi