Bridges solver enhancement. In the stage 3 solver, we were considering
authorsimon <simon@cda61777-01e9-0310-a592-d414129be87e>
Thu, 31 May 2012 18:10:11 +0000 (18:10 +0000)
committersimon <simon@cda61777-01e9-0310-a592-d414129be87e>
Thu, 31 May 2012 18:10:11 +0000 (18:10 +0000)
the possibility that an island might form an isolated subgraph by
connecting to one of its neighbours (and, if so, reducing the maximum
bridge count in that direction so that some bridge would have to go
elsewhere), but we were not also considering the possibility that it
might form an isolated subgraph by connecting to _more_ than one of
its neighbours. For instance, if you have a 3 adjacent to a 1, a 2 and
something else, then at least one bridge must go to the something-else.

Previously insoluble test case:
10x10m2:a2b4a5a2a2a1ga2d3b33a3a4c2aa3e1a22b2a4b4aa3b1a2b33a1e3aa2a1a2c23a3a3a4a2a

git-svn-id: svn://svn.tartarus.org/sgt/puzzles@9543 cda61777-01e9-0310-a592-d414129be87e


No differences found