Richard,
I created this killer:
JSudoku confirms that it has a unique solution. Then I gave it to SSolver(3.2.1) for scoring. After 30 minutes there was still no score.
After solving it on my own to confirm that it is possible to do it without T&E I started stepping through it using SSolver. Up to this point:
Code:
.-------------------------------.-------------------------------.-------------------------------.
| 67 49 28 | 2347 68 19 | 5 17 238 |
| 567 49 28 | 2347 35 19 | 46 17 238 |
| 57 1 3 | 47 56 28 | 46 9 28 |
:-------------------------------+-------------------------------+-------------------------------:
| 3 6 7 | 8 9 5 | 1 2 4 |
| 4 8 9 | 1 2 3 | 7 5 6 |
| 1 2 5 | 6 4 7 | 3 8 9 |
:-------------------------------+-------------------------------+-------------------------------:
| 28 7 1 | 23 38 4 | 9 6 5 |
| 9 3 6 | 5 17 28 | 28 4 17 |
| 28 5 4 | 9 17 6 | 28 3 17 |
'-------------------------------.-------------------------------.-------------------------------'
If I press F8 for a hint at this point nothing happens. I'm using the default scoring options. If I try to take another step I get this:
** Trying to set invalid zero candidate at r9c1My human approach to the killer at this point would be this:
97. 3 in r1c4 forces 8 in r1c5 -> problem placing candidates in r1c39 -> no 3 in r1c4
I did this step for SSolver. Now it started working again.
Now I have the same problem as before: F8 give me nothing, taking a step gives me this:
** Trying to set invalid zero candidate at r9c1My human approach is a short chain:
101. 7 in r1c1 -> r12c8 = [17], r12c6 = [91], r12c2 = [49], r1c45 = [26] -> problem, can't place 7 in r2c4 -> no 7 in r1c1
The rest could be done by SSolver.
I hope this information helps to to solve this mystery.
Cheers,
Nasenbaer