aboutsummaryrefslogtreecommitdiff
path: root/sort.c
diff options
context:
space:
mode:
authorSimon Tatham <anakin@pobox.com>2019-04-03 18:26:42 +0100
committerSimon Tatham <anakin@pobox.com>2019-04-04 23:58:31 +0100
commit453a2c1ca8971cb5b2cd1aef1c5c8a2fef6d107b (patch)
treec76643424ba24620b5774af3a3134a9f998822b9 /sort.c
parente2f52df5ec5a34ded78e95e9605c0bc810019bcd (diff)
downloadpuzzles-453a2c1ca8971cb5b2cd1aef1c5c8a2fef6d107b.zip
puzzles-453a2c1ca8971cb5b2cd1aef1c5c8a2fef6d107b.tar.gz
puzzles-453a2c1ca8971cb5b2cd1aef1c5c8a2fef6d107b.tar.bz2
puzzles-453a2c1ca8971cb5b2cd1aef1c5c8a2fef6d107b.tar.xz
Dominosa: allow set analysis even with adjacency.
I've always had the vague idea that the usual set analysis deduction goes wrong when there are two adjacent squares, because they might be opposite ends of the same domino and mess up the count. But I just realised that actually you can correct for that by adjusting the required count by one: if you have four 0 squares which between them can only be parts of 0-0, 0-1 and 0-2, then the only way this can work is if two of them are able to be the 0-0 - but in that case, you can still eliminate those dominoes from all placements elsewhere. So set analysis _can_ work in that situation; you just have to compensate for the possible double. (This enhanced form _might_ turn out to be something that needs promoting into a separate difficulty level, but for the moment, I'll try leaving it in Hard and seeing if that's OK.)
Diffstat (limited to 'sort.c')
0 files changed, 0 insertions, 0 deletions