source: trunk/third/diffutils/diagmeet.note @ 16149

Here is a comparison matrix which shows a case in which
it is possible for the forward and backward scan in `diag'
to meet along a nonzero length of diagonal simultaneous
(so that bdiag[d] and fdiag[d] are not equal)
even though there is no snake on that diagonal at the meeting point.


     85   1   1   1  159  1   1   17
        1   2   3   4
60
    1   2
1
    2       2   3   4
71
    3       3   4   5
85
    4   3   4   5
17
    5   4   5
1
    6       4   5   6
183
    7       5   6   7
10
    8       6   7
1
    9           6   7   8
12
                7   8   9  10
13
       10       8   9  10
14
           10   9  10
17
       10      10
1
   10   9  10
1
        8      10      10      10
183
    8   7       9       9       9
10
    7   6       8   9   8       8
1
    6   5               7       7
1
            5           6       6
1
                5       5       5
50
            5   4       4       4
1
                    4   3       3
85
            5   4   3   2       2
1
                            2   1
17
            5   4   3   2   1       1
1
                                1   0
     85   1   1   1  159  1   1  17