XnX (Cross-Number matriX) a
cross-number puzzle, similar to the Challenger
®
puzzle that appears in some newspapers, like San Francisco
Chronicle... (??? - Please, let me complete the formalities first.)
Accroding to Brian McBride, Challenger 'is a syndicated feature of King
Features Syndicate, Inc.
![King Features Sindicate Logo](licensingLogo.gif)
'.
You might ask how it happened that your humble servant, being among
those few coders who still live far away from CA and can't get used to
reading periodic literature online, knows something about SF Chronicle.
To be honest, I wouldn't have any idea about it, if not
Roderick
Young. It was my impression of his great PalmOS freeware named
X-num
, that triggered X#X development. Thank you Roderick, I owe
you so much.. Please forgive me that some ideas (hints, puzzle code,
etc) were shamelessly borrowed from X-num.
Another freeware, this time for PC, named
Challenger®
Puzzle Solver was written by
Brian McBride. Brian
attempts to explain his algorithm, and I will be hapy to find someone
who can clearly understand the explanations. To me the description
appears more concentrated on Java objects, than on the algorithm
itself. As far as I understood it, the algorithm checks all possible
combinations in a non-recursive way.
On the contrary, X#X solver (see
Appendix
) attempts to avoid testing all combinations, but is recursive which is
a disadvantage. Like probably most of you, I've read R. Sedgewick's
book and know that each depth-first algorithm allows a non-recursive
implementation, but I am out of time: X#X was not a part of my original
plan, while there are plenty other applications and updates in the
queue.
I regret that X#X is not a free application: after all we have to
live on that. But it is going to cost nearly as much as an average
lunch. An excellent chance to find out how many J2ME device owners
enjoy the application so much that can miss their lunch in order to buy
it :)
The development took about a fortnight (i.e. two US weeks). Most
of development work has been done in Linux 2.4.18 using Java 2 SDK v
1.4.2 (beta) and J2ME WTK versions 1.0.4_01 and 2.0 by
Sun
Microsystems Inc, ProGuard v 1.6 by
Eric Lafortune. I wish I
could say
'All' instead of
'Most of', but unfortunately
Motorola, Siemen, Ericsson and most of Nokia tools need for testing
wouldn't work anywhere apart from the 'smelly platform'.
The puzzle represents a square table where most of cells are left
blank, and the sums are specified for each row, column and both
diagonals. To solve the puzzle you need to place a digit from 1 to 9
in each blank spot so that all sums match.
Stardard Challenger
® puzzle has 4
rows and columns. This application allows to vary number of columns
from 3 to a maximum number that fits the screen.
You can either generate a puzzle, or enter your own puzzle. A
custom puzzle is checked for consistency (must have at least one
solution). The application can bring up to 500 solutions of a puzzle
(entered or generated). You may require a
single-solution puzzle
to be produced, in which case a minimum possible number of preset
elements will be added to a generated puzzle.
An unregistered copy has limitations, viz. time limit, can't solve
a custom puzzle.
The following releases are currently available: